University of Southern California Dissertations and Theses

Determinants Of Intercounty Migration: California, 1970-1973

(USC Thesis Other)

Determinants Of Intercounty Migration: California, 1970-1973

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content INFORMATION TO USERS This dissertation was produced from a microfilm copy of the original document. While the most advanced technological means to photograph and reproduce this document have been used, the quality is heavily dependent upon the quality of the original submitted. The following explanation o f techniques is provided to help you understand markings or patterns which may appear on this reproduction. 1. The sign or "target" for pages apparently lacking from the document photographed is "Missing Page{s)'\ If it was possible to obtain the missing page(s) or section, they are spliced into the film along with adjacent pages. This may have necessitated cutting thru an image and duplicating adjacent pages to insure you complete continuity. 2. When an image on the film is obliterated with a large round black mark, it is an indication that the photographer suspected that the copy may have moved during exposure and thus cause a blurred image. You will find a good image of the page in the adjacent frame. 3. When a map, drawing or chart, etc., was part of the material being ‘ photographed the photographer followed a definite method in "sectioning" the material. It is customary to begin photoing at the upper left hand corner of a large sheet and to continue photoing from left to right in equal sections with a small overlap. If necessary, sectioning is continued again — beginning below the first row and continuing on until complete. 4. The majority of users indicate that the textual content is of greatest value, however, a somewhat higher quality reproduction could be made from "photographs" if essential to the understanding of the dissertation. Silver prints of "photographs" may be ordered at additional charge by writing the Order Department, giving the catalog number, title, author and specific pages you wish reproduced. University Microfilms 300 North Zeeb Road Ann Arbor, Michigan 48106 A Xerox Education Company BERG, Dennis Floyd, 1940- DETERMINANTS OF INTERCOUNTY MIGRATION: CALIFORNIA, 1970-1973. University of Southern California, Ph.D., 1974 Sociology, demography U niversity M icrofilm s, A XEROX Company, A nn A rbor, M ichigan THIS DISSERTATION HAS BEEN MICROFILMED EXACTLY AS RECEIVED. DETERMINANTS OF INTERCOUNTY MIGRATION* CALIFORNIA, 1970-1973 by Dennis Floyd Berg A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (Sociology) June 1974 UNIVERSITY OF SOUTHERN CALIFORNIA T H E G RADUATE SC HO O L U N IV E R S IT Y PARK LOS A N G E LE S, C A L IF O R N IA 9 0 0 0 7 This dissertation, w ritten by under the direction of Dissertation Com mittee, and approved by a ll its members, has been presented to and accepted by T h e Graduate School, in p a rtia l fu lfillm e n t of requirements of the degree of P.e^i.s„Floyd Ber^ D O C T O R O F P H IL O S O P H Y ClluvdhDJf' £ Dean D ate DISSERTATION COMMITTEE Abstract This investigation evaluates the extent to which a "pull force" perspective of migration can be used to explain patterns of intrastate population redistribution. Using data derived from the California Department of Motor Vehicles* change of drivers license address file as a symptomatic indicator of migration, several determinants (destinational characteristics) of migration were employed to explain the intrastate distribution patterns of migrants in California for the period July 1, 1970 to July 1, 1973. Those determinants examined included population, migrant stock, distance, income, employment, commercial and industrial development and housing availability. The analysis consisted of estimating the degree to which the determinants in various combinations were able to explain the intrastate distribution of migrants from each of the fifty-eight counties in California. The conclusions drawn are based on sets of logarithmic multiple regressions (fifty-eight regressions per set). Migrant stock was eliminated from consideration after it was verified that its correlation with population (r) was .99. The remaining six determinants were found to explain an average of 80,0 percent of the variance in the intrastate distribution of migrants from the fifty-eight counties. The strongest variable in the model was popula- __2 _2 tion (fT = .632), followed by distance (0 = .193)i _2 housing (0 = .057) and commercial and industrial develop- _2 raent (0 = .041). Employment and income failed to make significant contributions to the explanation of migration _2 (0 's = .015 and .00? respectively). Removing population from the model decreases the proportion of variance explained by only .041. Of the five remaining determinants, commercial and industrial _2 development was found to be the strongest (0 = .243)» _2 _2 followed by housing (0 = .186) and distance (0 = .178). The independent contributions made by employment and _2 income approached zero (0 's = ,022 and .004 respectively). The fifty-eight counties in California were classi fied into four categories based on the magnitude of "gross intrastate migration" (GIM)» i.e., in-migrants from other California counties plus out-migrants to other California counties. Among the Metropolitan Growth Centers (those with GIM's of 25,000 to 225,000 licenses annually), which account for 75.4 percent of all intrastate migration, the six variable model accounted for 91.8 percent of the variance in migration. For the Novea (GIM's = 10,000 to 25,000) and Tertiary (GIM's = 6,000 to 12,000) Areas the iii figure was 90.4 percent and 88.6 percent respectively. Among the Rural Areas (those with GIM*s below 6,000) the model was able to explain only 67.3 percent of the variance in migration. With population excluded from the model it was found that the independent contribution of distance and housing varied directly with county type, while the independent contribution for commercial and industrial development varied inversely by county type. For the Metro, Novea and Tertiary Areas, housing was found to be the most important _2 _2 determinant (IF = .269), followed by distance (p = .239) _2 and commercial and industrial development (0 = .192), For the Rural Areas, however, commercial and industrial _2 development dominated (0 = .306), followed by distance _2 _2 (g = .156) and housing ({3 = .084). The results supported a "pull force" perspective of migration. The decided influence of county size (popula tion) as an attractive force was confirmed. The ability to partition population effects into dimensions was also demonstrated. Distance was shown to have a consistently inhibiting impact on the choice of destination. Finally, the data suggest systematic variations in the models' ability to explain migration and in the estimates of the relative importance of the variables in the model across county types, iv TABLE OF CONTENTS Chapter Page I THE IMPACT OF MIGRATION 1 Introduction .................................. 1 Defining the Problem ....... ........... 3 II THE DETERMINANTS OF MIGRATION 10 The Migration Experience ..... 10 A Review of the Literature............ 12 Proposed Model of Migration ......... .... 15 Distance and Migration............ 17 Size of Population and Migration ........... 20 Economic Determinants of Migration ........... 23 Growth and Migration ........................ 25 Migrant StocJc ........................... . 27 Interrelationships of Determinants ........... 30 III INTERSTATE AND INTRASTATE MIGRATION PATTERNS* CALIFORNIA, 1970-1973 33 The Department of Motor Vehicles' Change of Address F i l e ................. 33 Compatibility of the DMV Data ............ 34 Population and Gross Migration.......... 39 Basic Migratory Patterns* A Description . . . 42 IV THE DETERMINANTS OF MIGRATION* AN EMPIRICAL ANALYSIS 58 Empirical Indicators ..................... .. 5@ Method of Analysis . ................... 61 Intercorrelation Among Independent Variables . 63 Independent Variables and Migration ..... 67 Migrant Stock and Population................. 74- Model B . . . . . . . . . . . . . . . . . . . 80 Growth and Migration ................... 86 Model G ...................................... 93 Conclusions ................... ....... 98 v TABLE OF CONTENTS - continued Chapter Page V COUNTY DIFFERENTIALS 99 Introduction .......................... 99 County Types............................ . . . 100 Population, Distance and Migration ...... 105 Determinants of Migration by County Type . . • 112 VI SUMMARY AND CONCLUSIONS 120 County Types .................................. 123 General Patterns....................... 125 Conclusions................................ 126 BIBLIOGRAPHY 129 vi LIST OF TABLES Table 1 Estimates of Annual Net Migration for California Counties, 1970-1972............. 2 Intercorrelation Matrix for Estimates of Net Migration, Logarithmic Correlation Coefficients ............... 3 Logarithmic Correlation Coefficients between Population Size and Annual Out-migration for California Counties, 1970-1973 . . . . 4 Logarithmic Correlation Coefficients between Population Size and Annual In-migration for California Counties, 1970-1973 . . . . 5 Fifty Years of Growth in Selected Counties, California, 1920-1970 ........... ........ 6 Major Receiving Counties of Migrants Enter ing California from Out of State, 1970-1973 .................................. 7 Counties Reporting a Yearly Average of 10,000 or More Licenses Surrendered from Areas Outside of the County, by Origin of In-migrant, California, 1970-1973 . . . . . 8 Migratory Paths Exceeding a Gross Volume of 3.000 Licenses Per Year Between Califor nia Counties and Out of State, 1970-1973 9 Migratory Paths Exceeding a Gross Volume of 3.000 Licenses Per Year Between Two Northern Counties, California, 1970-1973 . 10 Migratory Paths Exceeding a Gross Volume of 3.000 Licenses Per Year Between Two Southern Counties, California, 1970-1973 • vii Page 37 39 41 41 43 44 46 48 50 52 LIST OP TABLES - continued Table 11 12 13 14 15 16 17 18 19 20 Migratory Paths Exceeding a Gross Volume of 3,000 Licenses Per Year which Connect a Northern County with a Southern County, California, 1970-1973 .............. . . . Zero Order Logarithmic Correlation Coefficients among Independent Variables Logarithmic Correlation Coefficients between Distance and Each of the Other Independent Variables by County................ . . . . Squared Logarithmic Correlation Coefficients between Each Independent Variable and Migration (M..), by County, California, 1970-1973 . .J............................. Intercorrelations (logarithmic) among Independent Variables, Reduced Model . , , p Proportion of Variance Explained (R ) in Migration from County i to County j by Population at j (P^), Migrant Stock at j (MS.) and Models J A thru D for Each J County i, California, 1970-1973 . . . Logarithmic Regression Coefficients and Coefficients of Multiple Determination of Migration (M. .), by County, California, 1970-1973 ................... Logarithmic Regression Coefficients and Coefficients of Multiple Determination of Migration (M. .), by County, California, 1970-1973 . . . V ......................... _2 Average Correlation Coefficients (R ) across Counties for Selected Models of Migration, California, 1970-1973 . . . .............. Logarithmic Regression Coefficients and Coefficients of Multiple Determination of Migration (M. •), by County, California, 1970-1973 . . . . ......................... gage 54 64 65 71 76 78 82 88 92 94 viii LIST OF TABLES - continued Table Page 21 Intrastate Migration Activity, by County, within County Types, California, 1970-1973 .................................. 101 22 Summary Statistics of the Interrelationships between the Variables Population, Distance and Migration, by County Type, California, 1970-1973 .................................. 106 _2 23 Average Multiple Correlation Coefficients (R ) for the Five Variable and Six Variable Models, by County Type, California, 1970-1973 .................................. 113 2k Average Squared Logarithmic Zero Order Cor relation Coefficients and Squared Multiple Regression Coefficients for Each Independent Variable with Migration (M..), by County Type, California, 1970-1973J.................115 LIST OF ILLUSTRATIONS Figure Page 1 Determinants of Migration .......... 16 2 Determinants of Migration* Suggested Model . . 31 3 Migratory Paths Exceeding a Gross Volume of 3,000 Licenses Annually Between California Counties and Out of Statei 1970-1973 . . . 49 4 Intercounty Migratory Paths Exceeding a Gross Volume of 3,000 Licenses Annually* Northern California, 1970-1973 ...... 51 5 Intercounty Migratory Paths Exceeding a Gross Volume of 3,000 Licenses Annually* Southern California, 1970-1973 ...... 53 6 Intercounty Migratory Paths Exceeding a Gross Volume of 3>000 Licenses Annually Connecting Northern and Southern Counties* California, 1970-1973 ..................... 55 7 California Counties as Classified by Annual Magnitude of Gross Intrastate Migration (GIM), 1970-1973 ......................... 103 x CHAPTER I THE IMPACT OF MIGRATION Introduction Each year, in the United States, about twenty percent of the population change their place of residence. Over a fourth of these moves are across county lines, moves which conform to the commonly accepted definition of migra tion. The current magnitude of migration taking place in the United States is producing extensive population redis tribution throughout the country. It has been estimated, for example, that enough moves are made in five years to depopulate all 210 U.S. metro politan areas with under one million residents and settle them anew somewhere else. At some point of aggregate effect, . . . , the exercise of the right to move impinges on the nature of the entire society. (Morrison, 1973* 1) Nowhere is the impact of migration more obvious than in the state of California. Over the past thirty years the population of California has increased by more than 13.1 million people. The greatest share of this increase (7.85 million or 59.8 percent) was due to the migration which took place from other states. Within the state, intercounty and interstate movements have combined to produce substantial changes in county populations. Over the last decade (1960-1970) the two counties representing the extremes of such changes were Trinity County and Orange County. In July, I960, the population of Trinity County, California was estimated to be 9»600. By July, 1970, it was found that the population had decreased to 7»700, a decline of 19.8 percent over the ten year period. During the same decade, the population of Orange County, California was estimated to have increased by 99.1 percent from a July, I960 figure of 719,500 to a July, 1970 figure of 1,432,400, In both cases, the crucial component of change was "net migration." During the decade of the 1960's, Trinity County experienced 600 more births than deaths (natural increase), a fact which under normal conditions would have suggested a growing population. During the same period of time, however, 2,500 more people migrated (moved) out of the county than migrated into the county. The net result of these two components, natural increase and net migration, was a decrease of 1,900 in the population. In contrast, Orange County experienced both a net natural increase of 164,600 (that many more births than deaths) and a net migration of 548,300 (that many more people migrating into 3 the county than out). The resulting effect was an increase of 712,900 in the population,1 While these examples represent the extremes of the changes which occurred in the populations of the counties in California during the 1960-1970 decade, they serve to illustrate well the role which the process of migration has assumed in influencing changes in the size of popula tions. The process of migration, with special emphasis on the determinants of migratory patterns, is the topic of this paper. Defining the Problem Population figures have for some time served as the basis for reapportioning legislative bodies so that con formity to the "one man, one vote" principle of democracy could be maintained. It was precisely for this reason that the decennial census of the United States was estab lished in 1880. More recently, however, population figures have begun to experience a wider role in the operation of governments. In California, population figures have been used for some years as the basis for the redistribution of state gas and cigarette tax monies to the various political Figures taken from California Population. 1971. published by the Population Research Unit, Department of Finance, Sacramento, California, May, 1972, p. 17. V 2 subdivisions of the state. More recently, the state legislature has enacted legislation which utilizes popula tion and population growth as one element in a set of mechanisms established to control tax rates (Assembly Bill 2008 and Senate Bill 90).3 State governments have not been the only agencies which have effectively increased the role which population plays in the operation of government. For several years now the federal government has been relying upon popula tion figures to distribute the benefits of federal help programs. Most noticeably among recent developments has been the "Omnibus Crime Control and Safe Streets Act of 1968" which allocates monies to states on the basis of population. The newly instituted federal revenue sharing program also utilizes population as one of several criteria for redistributing tax funds to various political entities (cities, counties and states) throughout the country (Public Law 92-512, February 22, 19?2). As the role of the size of the population of govern mental units has been growing in importance, so has the 2 See section #11005 of the California Revenue and Taxation Code and section #2107 of the California Streets and Highway Code. 3Briefly, SB 90 allows tax increases, without going before the voters, only to the degree that such increases combine with increases in assessed evaluations to produce an overall percent increase in revenues equal to the com bined percent increase in cost of living and the percent increase in population growth. role of the projected populations of those units. During the past year, the United States Environmental Protection Agency has adopted a set of stringent projections which are currently being used to restrict the federal funding of water treatment centers designed to meet needs exceeding those warranted by such projections. Other agencies, throughout national, state and local governments are beginning to place more reliance upon pro jections in their planning and budgeting processes. Recent comprehensive health facilities plans, for example, which are being submitted to the state for approval in conformity with recent federal legislation, rely almost entirely upon projected populations for their adopted recommendations concerning licensing of hospitals and types of hospital beds (Orange County Health Planning Council, 1973). While projections have been used for some time by business and industry in their planning process, only recently have they begun to assume a constricting role. Local governments have begun to evaluate long term bonded indebtedness more carefully, no longer can they assume that their populations will continue to grow or at least stay the same as when the indebtedness was incurred. Service agencies such as probation, welfare and health are becoming more accountable for their projected caseloads, the basis of which are found in the projected populations. Govern ment planning agencies have begun the processes of articulating growth policies (Davis and Styles, 1971). Citizen groups, developers, utility companies and planners alike are beginning to utilize conflicting projected populations in an effort to purse the consummation of often conflicting goals. Regional agencies, such as the Southern California Association of Governments and the Association of Bay Area Governments, have been federally authorized to coordinate the projection process in an effort to instill legitimacy to at least some sets of projected figures. Migration can produce substantial population redis tributions over short periods of time. Failure to anticipate these shifts can lead to the maldistribution of public facilities, severe strains on natural resources and the environment, and the loss of stability to the economic base of communities. Migration is an exceptionally complex phenomenon. Its implications as a problem for study are many and diverse. As Jansen (1969) has so articulately put iti Migration is a demographic problems it influences sizes of populations at origin and destinations it is an economic problems a majority of shifts in population are due to economic imbalances between areas* it may be a political problem* this is particularly so in international migra tions where restrictions and conditions apply to those wishing to cross a political boundary* it involves social psychology in so far as the migrant is involved in a process of decision-making before moving and that his personality may play an impor tant role in the success with which he integrates 7 into the host society* it is also a socio logical problem since the social structure and cultural system both of places of origin and of destination are affected by migration and in turn affect the migrant. . . (p. 60) The current lack of knowledge concerning the migra tion process severely restricts our ability to project and thus anticipate its impact. What are the factors which stimulate migration? If the magnitude of migration con- h , tinues as it has in the past* where will be the points of origin and where will be the points of destination? What will be the impact of such population redistribution upon the governments and people of those areas? The answers to these and other such questions continue to stand between us and our ability to anticipate shifts in the geographic distribution of the population. Such lack of ability has in the past and will continue to render the planning process at least ineffective and at worst, disas trous. As Greenwood and Sweetland (1972) have concludedi Since many of the current social, economic, and political problems of the metropolitan areas of the United States are caused by the pressure of population upon the resources available to the metropolitan areas, and since at least in part these problems are caused or intensified by it , "The percentage of total movers, mtracounty movers, intercounty movers and movers between states has remained fairly constant for the past 23 years. Total movers have fluctuated between 18# and 21# of the total U.S. population with the highs recorded in 1950» 1955» I960 and 196**. With the exception of intracounty movers, which indicate the greatest variation and have the heaviest influence, the remaining components of total movers have remained remar- kedly stable for the past two decades." (SCAG, 1973* P. 11) migration to or from such areas, policymakers should be particularly interested in the deter minants of migration to and from metropolitan areas, (p. 679) It is the intention of this research to continue the investigation of migration in an effort to add to the resolution of such questions as those posed above. The intended focus is on the determination and evaluation of those factors which appear to be most closely related to determining the magnitude and direction of intercounty migratory streams. While much effort has been expended on the investigation of international and interstate migra tion, little research has been devoted to intrastate move ments. In the following pages (Chapter II) a conceptual framework for approaching the study of intercounty migra tion will be developed. Included in this discussion will be a presentation of the "pull force" perspective of migration, the identification of those determinants thought to be most important to the explanation of migra tory behavior and the development of a general framework within which the determinants of migration are thought to fit. Following these discussions (Chapter III) will be a brief summary of the intercounty migratory patterns occurring within the state of California as evidenced by data derived from the California Department of Motor Vehicles' change of address file. Subsequent portions of the paper (Chapters IV through VI) will be devoted to the analysis of the propositions developed, the combined impact of the determinants upon migration and the county by county differentials in the model's ability to explain migration. CHAPTER II THE DETERMINANTS OF MIGRATION The Migration Experience Migration has been defined by Lee (1965) as "a per manent or semi-permanent change of residence. No restric tion is placed upon the distance of the move or upon the voluntary or involuntary nature of the act . . ." (p. 285). Normally, a change of residence is defined as migratory when the move involves the crossing of a specified system of boundaries. In the present case the concern is with a system of county boundaries, and thus, a change of resi dence from one county to another regardless of distance or reason becomes defined as migratory. Other commonly used systems of boundaries include Standard Metropolitan Statistical Areas, states and nations, Lee (1965) notes that no matter how short or how long, how easy or how difficult, every act of migration involves an origin, a destination, and an intervening set of obstacles. Among the set of intervening obstacles, we include the distance of the move as one that is always present, (p. 285) 10 11 Between any two areas i and j there is the potential for a flow of migrants to occur. Such a flow is composed of the movements from i to j, as well as the movements which take place from j to i. These two components have commonly been referred to as the migration stream and counterstream (Lee, 1965) of the flow or path. The directional movement having the larger magnitude of moves is normally denoted as the former. The balance of the stream and counterstream exchanges is known as "net" migra tion. The sum of the stream and the counterstream is known as "gross" migration. As a specific magnitude of net effect can be produced as a result of markedly differ ent grosses, recent years have seen the development of the concept of migration "efficiency," i.e., net migration divided by gross migration. These measures serve to parti tion the migration experience (flow or path) which takes place between any two locations i and j into its major component parts. Apart from these basic components of migration, each area of origin and destination is effected by a broader migration experience. Each area is involved in migratory exchanges with numerous other areas j. As a result of these exchanges, each location experiences in-migration (the sum of all flows into the area), out-migration (the sum of all flows out of the area), net migration (the balance of in- and out-migration), gross migration (the 12 sum of the in- and out-migrations), and even efficiency (net divided by gross). These measures can be applied either to describing the overall migration experience of a single location or to the individual flows which take place between any two areas. The latter, especially the stream and counterstream, form the most basic components of migration. Each of the other components which have been described is the product of streams and counter streams between areas. As such, the attention of this investigation will be focused on these latter aspects of migration, A Review of the Literature Concern with a theory of migration has been traced by Lee (1965) to the turn of the century and the work of E.G. Ravenstein (I885, 1889). From these early works came notions which continue to be evidenced in contemporary approaches to migration. Perhaps one of the most prominent examples is that of Ravenstein's seventh law as summarized by Lee in Ravenstein's (1889* 286) own words* (7) Dominance of the-economic motive* "Bad or oppressive laws, heavy taxation, anunattractive climate, uncongenial social surroundings and even compulsion (slave trade, transportation), all have produced and are still producing currents of migration, but none of these currents can compare in volume with that which arises from the desire inherent in most men to better them selves in material respects. ." (Lee, 1969* 283) 13 Ravenstein's assumption, that much of migration is due to man's desire to better himself, continues to be prominently employed in contemporary approaches to the topic. In one of the most recent attempts at formulating a theory of migration, Lee (1969) summarizes those "factors which enter into the decision to migrate and the process of migration ...” (p. 285). Lee's list includes* (1) Factors associated with the area of origin. (2) Factors associated with the area of desti nation. (3) Intervening obstacles, and (4) Personal Factors. (Lee, 1969* 285) In reviewing the research which has been conducted over the past twenty to thirty years, three other cate gories within which factors chosen for study commonly fall include 1 (5) Economic and labor market differentials which exist between an area of origin and an area of destination. (6) Intervening opportunities, and (7) Competing migrants. Most contemporary research concerned with explaining migration utilizes independent variables which fall into one or more of the above categories. Determinants repre sentative of each have been shown to be related to migra tion. Diverse rationales, varying approaches and a 14 general lack of data concerning migratory ’ behavior, however, have severely restricted coherence and synthesis in the field. Jackson (1969) has recently noted that while (t)he amount of empirical evidence available in the field of migration is enormous and the range and coverage of the statistical data is constantly improving . , . there has been only a relatively slight attempt to order the confusion with the development of theoretical propositions and models which would lend both elegance and under standing to this large and important subject, (p. 6) Noticeably missing from the evidence alluded to by Jackson, have been studies concerned with movement between areas smaller than SMSA's, Primarily dictated by the availability of data, the majority of the research reported to date has dealt with migration at the SMSA, state, regional or national levels. Caution concerning generalization of these findings to smaller levels has recently been voiced by Tarver and McLeod (1973)* . . . one cannot generalize these macro-level pre dictions to micro-units, such as cities, counties, state economic areas, or other small geographical areas, until appropriate tests are made. (p. 273) In this investigation, the selected determinants relate to the drawing power of the counties in the state of California. The variables chosen represent the dominant views currently evidenced in the literature. The decision to evaluate intercounty migration patterns from a "pull force" perspective was primarily stimulated by findings 15 reported by Alonso (1971) who noted, "that local conditions at the origin did not affect the rate of out-migration" (p. 6) and that "(c)ontrary to the no-push finding, pull- forces are very strong* local conditions at destination unequivocally affect flows" (p. 12), The units under analysis by Alonso were Standard Metropolitan Statistical Areas. Alonso's total effort was much broader than that pro posed here. Determinants which represented five of the seven categories noted on page 13 above (factors at origin, factors at destination, intervening obstacles, intervening opportunities and competing migrants) were evaluated. The local conditions (at destination) chosen for investigation coincided closely with those determinants currently receiving prominent attention in the literature. These included size of population, employment, income, migrant stock, climate and growth rate. With the^exception of climate'1 ' these variables, plus the universally considered determinant of distance, reflect that set of factors which has been chosen for study here. Proposed Model of Migration The specific elements which have been identified and selected to receive consideration here include* the 1Since this investigation is restricted to counties within the state of California, climatic variations were minimal and considered too small to be effective deter minants of within state intercounty migratory behavior. 16 distance between two counties i and j (D^)» the size of the population at county j (P^)* the size of the migrant v stock at county j (MS^)i the level of income in county j (1^)1 the level of employment in county j ' tlle level of industrial and commercial development taking place in county j (V.)t and the level of residential development J occurring in county j (H-). Each of these elements refers J to characteristics of the destinations which may be selected by the migrants from any single origin, including the distance which would have to be traveled to reach each destination. The model as initially proposed is of the form depicted in Figure 1. Figure 1 Determinants of Migration Dij 17 Distance and Migration Distance has been considered as a determinant by virtually every investigation attempting to explain migra tion, While there has been no consistent rationale employed with reference to its incorporation in explanatory models, it has generally been assumed that increasing dis tance tends to constrict the magnitude of migration which takes place between two locations, all other things being equal. This assumption, apart from its uncertain ration ale, has received repeated verification (Beals, Levy and Moses, 19671 Gallaway, et. al., 1967? Greenwood, 1969a, 1969b, 1970, 1971a, 1971b* Greenwood and Gormely, 1971* Greenwood and Sweetland, 1972* Lowry, 1966), In 19^6, Zipf introduced what has become known as the "gravity model" of migration. Borrowing from physics, distance was seen as a physical entity which constricted the gravity pull exerted by the size (population) of locations upon each other. As Lowry (1966) has put it, the "interaction between two places" (magnitude of migra tion) was seen as "a random event whose frequency depends on the number of potential actors and the distance or difficulty of interaction" (p. 8). As migration began to be viewed as a decision making process on the part of migrants* i.e., versus a simple physical phenomena* the rationale for incorporating dis tance into the model was redefined (Alonso, 1971* 18 Anderson, 1955* Miller, 1972* Stouffer, I960* Strodtbeck, 194-9). Distance began to assume the status of a proxy variable representing the costs involved in making a move. Such costs included not only the monetary expenses involved but also a set of nonmonetary costs experienced by the migrant. Greenwood and Sweetland (1972) explaini the nonmonetary costs of migration are psychic costs that involve the reluctance of an indivi dual to leave his family and friends and venture to unfamiliar surroundings, (p. 668) ........ The monetary component consists of two elements* (1) out-of-pocket transportation expenses . . .t and (2) the opportunity costs associated with migration, (p. 66?) The latter component of monetary costs refers to the "value of forgone alternatives that are ’sacrificed' during the move , , (p. 667). While working in a framework similar to the Zipf "gravity model," Stouffer (194-0) concluded that all other things being equal the magnitude of migration between two locations rather than being dependent upon the physical entity of distance was really a matter of the number of intervening opportunities which existed between the two points. This conclusion was verified by Stouffer through the substitution in the model of an independent measure of the number of intervening opportunities for that of distance. As a result, distance began to be treated as a proxy variable representing the possible extent to which such opportunities existed (Stouffer, I960* Isbell, 1944* 19 Strodtbeck, 194-9 and Anderson, 1955)• More recently Alonso (1971)» while retaining the concept of intervening opportunities, tended to discard distance as its legitimate operational definition. In its place Alonso used an inde pendent measure of opportunities based on population. The remaining major rationale which has been suggest ed, treats distance as an impingement to the flow of information. With the implicit assumption that only locations about which a potential migrant has information will be selected as a destination, the proposition that as distance increased, information availability decreased and thus the magnitude of migration between two locations would also decrease, was developed. In each of the aforementioned explanations, the rationale implies that increasing distances render migra tion between locations more difficult. In a recent article by Miller (1972) each of the above has been included under the single rubric of the "impediments to mobility." Distance is seen as the impediment because of its tendency to increase the cost of moves, to increase the differences between an origin and destination in terms of climate and local conditions, to increase the possi bility of the existence of intervening opportunities and to decrease the flow of information through social con tacts, newspapers, recruiters and job interviews (Miller, 1972* 475). 20 Regardless of the rationale chosen to explain its effect, the contention that distance does impede migration, as noted earlier, has received repeated verification (Alonso, 1971l Beals, Levy and Moses, 196?1 Fabricant, 19701 Greenwood, 1969a, 1969b, 19711 Greenwood and Sweetland, 19721 Karp and Kelly, 19711 Levy and Wadychi, 19721 Lowry, 19661 Sahota, 19681 and Tarver and McLeod, 1973, to name some recent examples). For the moment, and until that time that it becomes amenable to empirically distinguish the dimensions suggested in the above discus sion, distance will be viewed simply as an "impediment to mobility" and will serve as a general proxy variable assumed to contain confounding effects of each of the above suggested mechanisms. As such it is hypothesized that t As the distance of alternative points of destination from a point of origin increases the magnitude of migration will decrease, all else being equal. Size of Population and Migration Since the original formulation of the "gravity model" (migration as a product of the forces of population mediated by distance) by Zipf in 19^6, the size of the population both at origin and destination has received continued attention from those studying migration. While receiving continued verification as a major determinant of migration, its theoretical position remains obscured. 21 In 19?1» for example, Karp and Kelly state* Population of the destination city is probably the least pure (both theoretically and opera tionally) of the variables used in this study. While past studies of trends in migration show that the size of a metropolitan area acts as a force of attraction for migrants, it is diffi cult to specify exactly what this attraction consists of. (p. 21) While the difficulty of clearly specifying the role of population size in migration is apparent there have been several suggestions pertaining to its underlying influ ences. Zipf, for example, saw the influence of population as conforming to Newton's Law of Gravity where the "inter action between two places is a random event whose frequency depends on the number of potential actors ..." (Lowry, 1966* 8). Karp and Kelly provide as one of their defenses for considering population as a determinant of migration the fact that "it is possible to state some of the attractive factors that a large population implies . ." (1971* 21). Included in their discussion were such factors as specialized services, entertainment by famous stars, big league sporting events and other such attractions most frequently found in the larger metropolitan areas (Karp and Kelly, 1971* 21). William Alonso (1971) has reported that "the number of migrants leaving an area is slightly less than propor tional to the local population" (p. 10). He continues, 22 (i)t is almost as if an urban area were a radio active body, emitting particles at a steady rate regardless of such local conditions as heat or vibration. The number of particles is slightly less than proportional to the size of the body because, with increasing size, an increasing proportion of the particles is trapped within the body on their way out, (p. 10) Alonso (1971) also found "the number of migrants arriving at the destination, , . . , is almost proportional to the population size . , (p, 11), The reason sugges ted in this case, unlike the previous alluded to physical science mechanism, is that population at the point of des tination be considered "as a measure of the number of opportunities available to the migrant at that location . , ." (p, 11). A similar logic has been employed by Greenwood (1970) who reasoned that "the greater the population of the destination state, the greater is likely to be the number of expected job opportunities" (p, 381), The influence of population on migration has received consistent verification. Employing the broad assumption that migration is an option open to those seeking to better themselves and given the fact that with increased size comes increased division of labor and thus greater diversi fication of specializations and greater labor force demands, it appears reasonable that movement would be toward populated areas. It is consequently hypothesized that the magnitude of streams toward alternate destinations will vary directly with the size of destination populations. 23 Economic Determinants of Migration A consistently held contention has been that economic factors of competing areas influence the migrants' choice of destination. Receiving major consideration from labor market migration theorists the predominant determinants considered have been employment and income. While the evidence accumulated to date has not been extremely impressive, further consideration of these factors appears warranted. In 1965# Tarver failed "to establish any significant relationship between unemployment and migration rates after" (p, 220) adjustments for other variables were made. Likewise, it was found that mean "family income" had "no perceptible influence upon the movement of , , , residents between counties" (pp. 220-221), In explaining these findings Tarver noted that "income varies directly with population size" (p, 220) and thus the possibility that its effect was absorbed by the population variable in his model, Lowry (1966) found that "migration from place i to place j was encouraged by high wages at j , , ," (p, 22), Gallaway, Gilbert and Smith (1968) hold that "income differences are a significant determinant of interstate movements of population" (p. 244), Greenwood and Sweetland (1972), however, while noting that , , , "migrants do indeed tend to move away from low-income Zk SMSA’s and toward high-income SMSA's " (pp. 671-672) report that "migrants typically have no significant tendency to move to high-income SMSA's" per se (p. 675). Sjaastad (1961), Fabricant (1970), Karp and Kelly (1971) and Tarver and McLeod (1973) all report similar findings. It should be noted that these findings refer to those situ ations in which "gross" migration served as the dependent variable and do not necessarily apply to those instances where the major interest is "net" migration. Even though the findings to date are not that impressive, there appears to be enough conflicting results to support the continued inclusion of such factors in migration models. Excessive unemployment and/or low income potential appear to be logical deterrents to the selection of a point of destination. While the term excessive is suggestive of a threshold effect of these factors, and indeed would appear to be a reasonable course of investigation, it remains beyond the scope of the present endeavor. From the range of economic determinants which have occurred in reported research the two most prominently evaluated determinants have been selected for inclusion in the present models i.e., employment and income. It is hypothesized that the magnitude of migratory streams will vary directly with the rate of employment, and secondly 25 that the magnitude of migratory streams will vary directly with the overall level of income of alternate points of destination. Growth and Migration Evaluation of growth itself as a determinant of migration has as yet received much attention from those interested in explaining migration. The first major suggestion of its impact came in 1970 from Alonso and Medrich. Concerned with the development of a national urbanization policy, Alonso and Medrich identified "two varieties of growth centers. Induced growth centers are those in which public policy is trying to promote growth" while "spontaneous growth centers are those that are growing without benefit of conscious or explicit policy" (p. 2). In evaluating the role of Spontaneous Growth Centers in the urbanization of the United States, Alonso and Medrich found that* Since the beginning of the century (and presum ably earlier) a very large share of American metropolitan growth, and a far larger share of the net inmigration into metropolitan areas, has been absorbed by those metropolises which grew substantially faster than the metropolitan set. This share has been increasing recently, in spite of the declining importance of metro politan inmigration, as a result of a more active and selective inter-metropolitan migration. As the number of areas with substantial net inmigra tion has increased, so has the number of metro polises which are net exporters of people, (p. 27) 26 The model established by Alonso in 1971 suggested that "(o)ther things being equal , . . people head towards places that are growing fast" (p. 13). While the inclu sion of the growth at destination variable by Alonso was in the interest of maintaining symmetry between origin and destination factors which were being considered, the results appeared to be surprising and (by the way) signi ficant. Two interpretations were offered* The first is that migrants regard not only the present population size of a place as indicating the number of opportunities available there, but that they also look at the growth rate as an indicator of the number of opportunities being generated. The second interpretation is based on the effect of informal networks where by early immigrants sent the glad news back home and encourage their friends and relatives toward the same destination. (p. 13) The recent upsurge of interest in the concept of migrant stock, that being alluded to in the latter interpretation, provides the mechanism for separating these apparently distinct effects. As a result, the perspective of growth as the development of opportunities is the sole approach supported here. Two distinct growth factors have been selected for consideration. These are industrial and commercial development and housing availability. While the former more typically conforms to the opportunities dimension of growth, the latter is felt to represent a second and until now generally neglected aspect of development. Housing 27 development is symptomatic of growth and adds an added dimension to the potential impact of a growth in oppor tunities upon migration. It is anticipated that the magnitude of migratory flows to alternate destinations will vary directly with the amount of industrial and commercial development taking place at the points of destination. And secondly it is hypothesized that the magnitude of migratory flows to alternate destinations will vary directly with the availability of housing existing in each destination. Migrant Stock Perhaps the migration determinant receiving the most enthusiastic attention in recent years has been migrant stock. The concept as employed here is intended to refer to those who have previously migrated to an area and not as Lianos (1972) has used itj i.e., to refer to the number of people who are willing to migrate. The rationales as developed to date seem to suggest two conflicting effects of an accumulating migrant stock. First there are indications that aB the migrant stock of an area increases, information about that area becomes more rapidly and widely disseminated, thus increasing the tendency for that area to be selected as a destination. This conforms to the interpretation offered by Alonso (1971* 13). On the other hand, it would appear that as the magnitude of the migrant stock increased, changes in the attractiveness of the area would occur which would reduce its potential drawing-power and thus reduce the tendency to select the area as a destination. To date there is no evidence to suggest that the latter interpre tation has any validity, therefore the focus will be on the ability of migrant stock to attract more migration. Greenwood (1970) not only suggests that more infor mation concerning employment and opportunities is likely to be disseminated out of an area with increasing migrant stock, but also indicates that the presence of friends and relatives helps social transitions and thus reduces potential economic and noneconomic costs involved in con summating a change of residence. In his findings, Greenwood reports that "failure to include the migrant stock variable in the estimated relationship causes the true direct effect of most other variables to be obscured" (p. 189)1 i.e., if "migrant stock is not taken into account the true direct relationship between different variables and migration is overstated" (p. 190). While the utility of the migrant stock variable has received continued support (Alonso, 1971* Fabricant, 1970* Greenwood, 1969a, 1970* Hagerstrand, 1957* Larber, 1972* Nelson, 1959* and Tarver and McLeod, 1973)» Nelson (1959) has suggested and Greenwood (1970) has reiterated its critical relationship with the variable of population. 29 . . . population variables will pick.up the effects of the migrant stock variable if the migrant stock variable is not itself included in the estimated relationship. The reason for this is that the larger is the population of state j, the more likely is state j to have received a greater number of migrants from any given state in the past. (Greenwood» 1970i 383) This interpretation would appear not only to be legitimate for states but for counties as well. The larger the county, the larger the migrant stock population would appear to be easily confirmed. While this discussion might suggest that the need for including the migrant stock variable is negligible given that the size of the population appears in the model, the question is being left open for empirical resolution. It should be pointed out, however, that there are reservations concerning the data available for testing this hypothesis. The evaluation of movements from particular origins (i) to given destinations (j) would suggest that the migrant stock be that quantity of people who currently reside in j who had previously resided in i. While this information has in the past been made available and utilized by some for movements between SMSA's and states, no such set of data exists which relates to previous movements between the areas under study. Consequently, a lack of correspondence of the measurement with the concept as defined is unavoidable. 30 The expectation is that as migrant stock increases in an area the tendency to select that area as a desti nation will also increase. Interrelationships of Determinants It is realized that each of the elements which have been discussed does not operate independently of each other in influencing the direction and magnitude of migra tory flows. It has already been noted, for example, that the size of population depends heavily upon the magnitude of previous migration. It is therefore anticipated that migrant stock will be highly correlated with population size. Population also impinges upon some of the other determinants under consideration. Prehn (1967), for example, found a direct association between size of place and median income. Industrial and commercial development is also expected to be directly associated with population size. As development within the areas of industry and commerce occurs, increased employment can be anticipated and thus increased activity in the area of housing construction. Each of these determinants (population, migrant stock, employment, income, commercial and industrial development and housing availability) in turn adds to the attractiveness or pulling-force of an area competing among 31 other alternate destinations for the prospective migrants from any particular point of origin. The major mitigating force impinging upon such pull-forces is the distance which would have to he traveled to reach each of the points of destination. Figure 2 Determinants of Migration! Suggested Model MS. 1 +) (+1 (+)j A recapitulation of the model just discussed is represented in Figure 2 with the appropriate direction of anticipated relationships indicated. Prior to the analysis of the implications of this model, a description of the interstate and intrastate migratory experiences of the fifty-eight counties in the state of California will be presented (Chapter III), CHAPTER III INTERSTATE AND INTRASTATE MIGRATION PATTERNS: CALIFORNIA, 1970-1973 The Department of Motor Vehicles* Change of Address File The migration data used in this study has been secured through the Population Research Unit of the State of California Department of Finance from the California Department of Motor Vehicles (DMV). Since 1969» the California DMV has been maintaining a change of address file containing the records of all reported changes in address of all persons holding California drivers licenses or California I.D, cards (issued to nondrivers through the DMV). The file also contains the previous addresses of those who surrendered their out of state licenses for a California license and a record of the addresses of those with California licenses who have surrendered their licenses to an out of state jurisdiction. Data from the change of address file are summarized in a 60 by 60 matrix form with the sixty coordinate 33 31* identifiers consisting of the fifty-eight counties of the state, an out of state category and an APO (overseas) category. On a monthly accumulative basis the matrix includes the number of licenses surrendered from each of the sixty possible points of origin to each of the sixty points of destination. These changes are categorized by the age (four categories* under 25 years of age, 25-^4 years of age, years and 65 and older) and sex of the license holder. Compatibility of the DMV Data In an effort to assess the ability of the DMV change of address data to adequately represent California migra tory behavior, an analysis of the file’s compatibility with other indicators of migration was conducted. Since the ability to disaggregate the migration phenomena into streams is unique to the DMV file, assessment of the file’s compatibility with other sources was based on estimates of net migration. The four independent estimates of the magnitude of annual net migration occurring in each county for the period April 1, 1970 to April 1, 1972 were* (1) Department of Motor Vehicles (DMV) net address changes* A yearly average of net license changes of address for the period July 1, 1970 to July 1, 1972 was computed for each of the fifty-eight counties in the state by subtracting the two year yearly average of movements 35 into the county from other counties in the state, from other states and from APO addresses, from the two year yearly average of movements out of the counties to other counties, to other states and to APO addresses. (2) California State Department of Finance (DOF) residual net migrationi The Department of Finance has compiled population estimates for each county in the state as of July 1, 1972, Also estimated is the magnitude of natural increase and net migration as found through the residual method (P^ - ” NIt-l to t = net migra’ ti°n)« This two year net migration figure was divided by two pro viding a yearly average for the two year period compatible with the DMV computation. (3) United States Census Bureau (USCB) Method II net migration estimates* In the cooperative federal-state program of county estimates, the United States Census Bureau estimates county population using Method 11.^ These computations yield an estimate of the net migration which has taken place since the last census. Using the April 1, 1972 estimated net migration and dividing by two ■^Method II estimates net migration by* a) comparing current school enrollments in each county with that which would be expected based on survival rates applied to the latest census count 1 b) adjusting these rates for the "school age/under 65" migrant ratio of the previous decade 1 c) applying the adjusted rate to compute the number of migrants under 65 and d) adding to these the number of migrants 65 and over as estimated from Medicare statistics (Department of Commerce, Bureau of the Census, 1966). 36 a yearly average net migration for each county was computed roughly corresponding in time to the estimates made with the above sources of data. (*0 United States Internal Revenue Service (IRS) estimates of net migration* A one percent sampling of tax returns is made each year from which estimates of net migration are compiled. The estimates for the years 1970-1971 and 1971-1972 were averaged to produce an esti mated yearly average net migration for each county. The results of the above computations are reported in Table 1. It should be noted that neither the DMV nor the IRS data should be of the order of magnitude of the DOF and USCB estimates. The latter are estimates of the total number of people, while the former are estimates of changes of address on drivers licenses and tax returns respectively. The conversion of the latter two estimates into people equivalents has as yet to be accomplished. Both are relatively new attempts at tracking migration and the process of establishing conversion factors requires considerable time and resources. Each of the estimates of net migration was converted to logarithms. Logarithmic transformations will be used throughout this analysis to correct for the distortion in the linear assumption caused by the extensive range in values evident on most variables being evaluated (Blalock, 1972*^09). The logarithmic correlation coefficients 37 Table 1. Estimates of Annual Net Migration for California Counties, 19?0-19?2 Counties D0Fa USCBb IRSC DMVd Alameda 900 5,026 - 2,900 - 1.324 Alpine 50 85 0 8 Amador 250 542 300 403 Butte 2,100 2,650 2,400 1.030 Calaveras 400 603 700 404 Colusa 250 99 400 98 Contra Costa 6,650 - 1,637 2,150 1,953 Del Norte 300 249 150 156 El Dorado 2,300 2,042 1,500 1.592 Fresno 2,250 3.277 1,400 135 Glenn 100 267 100 1 Humboldt - 1,050 - 113 1,100 351 Imperial 300 366 200 - 295 Inyo 200 50 100 421 Kern 700 203 200 400 Kings 400 - 477 100 116 Lake 1,000 1,444 850 1,110 Lassen 50 281 0 354 Los Angeles -108,550 -77.044 -59,650 -64,454 Madera 550 586 500 89 Marin - 1,000 - 1,064 100 2,738 Mariposa 500 367 200 309 Mendocino 0 921 1,000 846 Merced 1,050 950 350 414 Modoc 200 299 300 68 Mono 550 384 50 349 Monterey 200 3,098 650 1,891 Napa 1.950 1,265 450 746 Nevada 650 931 450 538 Orange 49,350 26,439 22,850 22,131 Placer 2,250 2,637 900 1.551 Plumas 350 241 700 300 Riverside 8,850 7,556 4,750 5.724 Sacramento 8,750 9.259 1,300 188 San Benito 250 15 300 55 San Bernardino 850 - 3.100 2,500 1,464 San Diego 33.000 35,784 9,800 19,064 San Francisco -16,100 -13,345 -12,100 - 5.253 San Joaquin 1.650 5 100 - 1.305 San Luis Obispo 1.250 3.597 700 2,719 San Mateo - 2,700 841 - 3.750 - 3.092 Santa Barbara 100 2,255 300 760 Santa Clara 19,000 13.943 4,550 6,292 Santa Cruz 4,650 5.289 3,450 3,814 Shasta 1,000 308 1.250 264 38 Table 1 — Continued Counties DOF USCB IRS DMV Sierra 0 62 100 34 Siskiyou 350 354 0 566 Solano 1,950 3.083 800 133 Sonoma 5,200 4,934 4,750 4,253 Stanislaus 3,750 1,376 950 791 Sutter 550 202 - 350 - 16? Tehama 700 177 100 82 Trinity 400 508 350 283 Tulare 2,350 2,044 800 464 Tuolumne 550 642 550 782 Ventura 9,750 1,454 5,600 4,844 Yolo 1,000 1,744 1,000 935 Yuba 200 623 650 75 p California State Department of Finance (units = number of people). ^United States Bureau of the Census (units = number of people). cInternal Revenue Service (units = income tax returns). California Department of Motor Vehicles (units = driver licenses). 39 between each pair of estimates were computed. The results of these computations are reported in Table 2, While discrepancies between the estimates of net migration exist (Table 1), the overall intercorrelation of the four indicators is high (Table 2). Table 2, Intercorrelation Matrix for Estimates of Net Migration, Logarithmic Correlation Coefficients DOF USCB IRS DMV DOP 1.000 .978 .982 .986 USCB 1,000 .948 .972 IRS 1.000 .984 DMV 1.000 Population and Gross Migration Recalling the assertions of Alonso (1971) that "the number of migrants leaving an area is slightly less than proportional to the local population" (p. 10) and that "the number of migrants arriving at the destination . . . is almost proportional to the population size . , . (p. 11) an evaluation was made of the correlation between size of population and gross out- and gross in-migration. The log of the total number of licenses leaving each county each year as well as the log of the number of licenses entering each county were correlated with the log of the population of each county at the beginning of each of the three yearly periods. The population at each period was also correlated with all activity occurring in following periods to provide an indication of the extent to which lagged effects operate. The results of these logarithmic regressions are presented in Tables 3 and k. Size of population was found to be a strong indicator of both the magnitude of gross out-migration and the magni tude of gross in-migration. Overall, the coefficients are consistently somewhat smaller and the standard errors of estimate somewhat larger for the relationship between population and in-migration when compared with the rela tionship between population and out-migration. While lagged effects appear trended for in-migration, they are virtually nonexistent in the out-migration regressions. These results generally confirm Alonso's observations. 41 Table 3. Logarithmic Correlation Coefficients between Population Size and Annual Out-migration for California Counties, 1970-1973 Licenses Out Estimated Populations 7/1/70 7/1/71 7/1/72 7/1/70-7/1/71 .994 (.202) - 7/1/71-7/1/72 .995 (.173) .995 (.171) 7/1/72-7/1/73 .994 (.183) .995 (.180) .995 (.173) Figures in the table are logarithmic correlation coefficients with the computed standard error in parentheses. Table 4. Logarithmic Correlation Coefficients between Population Size and Annual In-migration for California Counties, 1970-1973 Estimated Populations Licenses In 7/1/70 7/1/71 7/1/72 7/1/70-7/1/71 .985 (.294) - - 7/1/71-7/1/72 .984 (.289) .985 (.278) - 7/1/72-7/1/73 .977 (.338) .978 (.330) .979 (.322) Figures in the table are logarithmic correlation coefficients with the computed standard error in parentheses. 42 Basic Migratory Patternsi A Description Fifty Years of California Growth* 1920-1970 Between April 1, 1920 and April 1, 1970 the popula tion of California increased by 16.5 million. Over 80 percent of that growth occurred in twelve of the state's fifty-eight counties. Over 52 percent occurred in the three southern counties of Los Angeles, Orange and San Diego alone (see Table 5)* An analysis of the out of state drivers licenses which have been surrendered in California for the period July 1, 1970 to July 1, 1973 (out of state in-migrants) indicates that 83.1 percent of all migrants moving into the state have chosen one of these same twelve counties as their destination (see Table 6), It should also be noted that the three southern counties of Los Angeles, Orange and San Diego serve as destinations which have been selected by 51 percent of these out of state in-migrants. Reported in Table 6 for the twelve counties involved in the major out of state exchanges is the net out of state effect, i.e., the number of drivers licenses received by each county from out of state minus the total number of licenses lost by each county to out of state. Seven of the twelve counties are maintaining a positive balance from out of state exchanges, while five of the counties appear to be areas from which returning migrants are Table 5. Fifty Years of Growth in Selected Counties, California, 1920-1970 Counties Population (in 1,000's) 4/l/20 4/1/70 Numerical Increase Percent Increase Percent of State Increase Los Angeles 936.5 7040.3 6103.8 651.8 36.90 Orange 61.4 1420.7 1359.3 2213.8 8.22 San Diego 112.2 1357.9 1245.7 1110.3 7.53 Santa Clara 100.7 1066,2 965.5 958.8 5.48 Alameda 344.2 1073.2 729.0 211.8 4.41 San Bernardino 73.4 682.2 608.8 829.4 3.68 Sacramento 91.0 634.2 543.2 596.9 3.28 San Mateo 36.8 556.6 519.8 1412,5 3.14 Contra Costa 53.9 555.8 501.9 931.2 3.03 Riverside 50.3 459.1 408,8 812.7 2.47 Ventura 28.7 378.5 349.8 1218.8 2.11 San Francisco 506.7 715.7 209.0 41.3 1.26 Totals 2395.8 15940.4 13544.6 565.3 81.88 California Total 3426.9 19968.0 16541.1 482.7 100.00 - ? = ■ VjJ 44 Table 6. Major Receiving Counties of Migrants Entering California from Out of State, 1970-1973 Percent of the Yearly Average Total Out of Number of Licenses State Licenses Surrendered from Surrendered Out of State in California Southern Counties Los Angeles 72,700 31.0 San Diego 26,750 11.4 Orange 20,030 8.5 San Bernardino 7,480 3.2 Riverside 5,680 2.4 Ventura 4,580 2.0 Total 134,220 58.4 Northern Counties Santa Clara 15*150 6.5 Alameda 12,820 5.5 San Francisco 11.670 5.0 San Mateo 6,380 2.7 Sacramento 6,370 2.7 Contra Costa 5,660 2.4 Total 58,050 24.7 Major Counties Total 195.270 83.1 45 leaving the state. The major receiving counties in both the south and north are retaining a positive net out of state balance while all but one of the minor receiving counties (San Mateo) have a negative net out of state balance. In a broader look at the selection of California counties as points of destination, Table 7 presents all counties which have reported a yearly average of 10,000 or more licenses surrendered (or changes of address) to it from areas outside of the county. These transactions are reported by origin of the in-migrants, i.e., either as from other counties in the state or from other states. The nineteen counties (32.7 percent of the fifty-eight counties) served as points of destination for 77 percent of all movement to or within the state. While most of these major points of destination are receiving the predominance of their in-migrants from other counties in the state (fifteen of the nineteen receiving approxi mately 70 percent or more of their in-migrants from other counties), four show a noticeable dependence upon out of state traffic. These include Los Angeles which received 50.3 percent of its in-migrants from out of state, San Diego (42.6 percent), San Francisco (38*4 percent) and Monterey (34.4 percent) counties. Finally, an analysis of the major migratory paths involving California counties was conducted. Those Table 7. Counties Reporting a Yearly Average of 10,000 or More Licenses Surrendered from Areas Outside of the County, by Origin of Inmigrant, California, 1970-1973 Number of Licenses Surrendered to County Per Year From Other From Other Total Counties States Counties N Percent N Percent N Percent Los Angeles 144,597 100.0 71,898 49.7 72,699 50.3 Orange 86,593 100,0 66,563 76.9 20,030 23.1 San Diego 62,715 100.0 35,974 57.4 26,741 42.6 Santa Clara *>9.981 100.0 34,827 69.7 15.15*> 30.3 Alameda 42,661 100.0 29,840 69.9 12,821 30.1 San Bernardino 32,095 100.0 24,612 76.7 7,483 23.3 San Francisco 30,424 100.0 18,755 61.6 11,669 38.4 San Mateo 28,538 100.0 22,159 77.6 6,379 22.4 Riverside 27,782 100,0 22,105 79.6 5,677 20.4 Sacramento 26,839 100.0 20,465 76.3 6,374 23.7 Contra Costa 26,382 100.0 20,725 78.6 5.657 21.4 Ventura 21,729 100.0 17.154 78.9 4,575 21.1 Santa Barbara 14,669 100.0 10,424 71.1 4,245 28.9 Fresno 13,335 100.0 10,907 81.8 2,428 18.2 Marin 12,882 100.0 9,031 70.1 3,851 29.9 Sonoma 12,722 100.0 10,943 86.0 1,779 14.0 Kern 12,023 100.0 9,352 77.8 2,671 22.2 Monterey 11.970 100.0 7,849 65.6 4,121 34.4 Santa Cruz 10,095 100.0 8,686 86.0 1,409 14.0 - p - o\ selected and which appear in Tables 8 through 11 include all paths between two counties or between a California county and out of state which were found to have a gross (licenses in plus licenses out) migration of 3,000 or more licenses per year)• The flow systems described in Tables 8 through 11 are depicted in Figures 3 through 6. There are twenty counties in California which exper ience an exchange of migrants with other states greater than 3.000 licenses per year (Table 8 and Figure 3). Of these, eleven (55 percent) are experiencing a positive net balance from such exchanges. These positive net balances range from a high of 8,709 for San Diego County to a low of 323 for Solano County. Those experiencing a negative net balance with other states range from a -1,527 net for San Bernardino to a -200 net for Fresno (see Table 8). The within state circulation system of migrants breaks into three categoriesj i.e., those movements among northern counties (Table 9 and Figure 4), those movements among southern counties (Table 10 and Figure 5) and finally those paths between southern counties and northern counties (Table 11 and Figure 6). The magnitude of movements between counties in the southern part of the state is almost two and a half times that of the magnitude of movements between counties in the northern part (182,797 versus 74,863). The net effect of the single Los Angeles to Orange stream of 30,045 per 48 Table 8. Migratory Paths Exceeding Licenses Per Year between Out of State, 1970-1973 a Gross Volume of 3,000 California Counties and County Grossa Netb Efficiency0 Los Angeles 141,609 3.787 2.7 San Diego 44,773 8,709 19.5 Orange 38,950 1,110 2.9 Santa Clara 27,523 2,784 10.1 Alameda 23,507 2.134 9.1 San Francisco 17,186 6,152 35.8 San Bernardino 16,492 - 1.527 9.3 Sacramento 13,142 - 393 3.0 San Mateo 11.959 800 6.7 Riverside 11,720 ~ 367 3.1 Contra Costa 11,610 — 296 2.6 Ventura 9,682 - 531 5.5 Santa Barbara 8.158 331 4.1 Monterey 6,907 1,335 19.3 Marin 6,348 1.354 21.3 Kern 5,966 - 624 10.5 Solano 5,625 323 5.7 Fresno 5,060 200 4.0 Sonoma 3.800 — 242 6.4 San Joaquin 3.722 - 53^ 14.2 d The number of licenses from out of state surrendered in each county plus the number of licenses from the county surrendered to an out of state jurisdiction. bNet effect to countyi i.e., the number surrendered to county minus the number surrendered by the county to out of state. °The net divided by the gross multiplied by 100. *tOrv < SEs^ h, ' S p. F , G < J * e , Z ? J S > A G/?o. Or J9?3> c °o«tZ 00 Table 9. Migratory Paths Exceeding a Gross Volume of 3,000 Licenses Per Year between Two Northern Counties, California, 1970-1973 Paths Streama Counter?- Stream Gross0 Netd Efficiency® Alameda to Contra Costa 7502 5474 12976 2028 15.63 San Francisco to San Mateo 8727 3791 12518 4936 39.43 San Mateo to Santa Clara 8173 4326 12499 3847 30.78 Alameda to Santa Clara 4013 3192 7205 821 11.39 San Francisco to Alameda 3845 2556 6401 1289 20,14 San Francisco to Marin 3784 1528 5316 2260 42.50 Santa Clara to Santa Cruz 2706 1254 3960 1452 36.67 San Francisco to Santa Clara 2276 1555 3831 721 18,82 San Mateo to Alameda 2242 1468 3710 774 20.86 Placer to Sacramento 1897 1327 3224 57 0 17.68 Yolo to Sacramento 1669 1554 3223 115 3.57 aThe number of licenses from the first county surrendered in the second county. fcThe number of licenses from the second county surrendered in the first county. cThe sum of the stream and the counterstream. The difference between the stream and the counterstream. eThe net divided by the gross multiplied by 100. 51 FIGURE 4 INTERCOUNTY MIGRATORY PATHS EXCEEDING A GROSS VOLUME OF 3000 LICENSES ANNUALLY: NORTHERN CALIFORNIA, 1970 - 1973 p la c e * LEGEND G —CROSS LICENSES ™ L0 F N — NET LICENSES S V, M - STREAM DIRECTION DIDTN PROPORTIONAL TO NET G=SS1S CONTRA COSTA Table 10, Migratory Paths Exceeding a Gross Volume of 3,000 Licenses Per Year between Two Southern Counties, California, 1970-1973a Paths Stream Counter- Stream Gross Net Efficiency Los Angeles to Orange 48726 18681 67407 30045 44.57 Los Angeles to San Diego 14834 7272 22106 7562 3^.21 Los Angeles to San Bernardino 12668 7591 20259 5077 25.06 Los Angeles to Ventura 10814 5104 15918 5710 35.87 Los Angeles to Riverside 8833 3654 12537 5229 41.71 Orange to San Diego 4523 3337 7860 1186 15.09 San Bernardino to Riverside 3711 3337 7048 374 5.31 Los Angeles to Santa Barbara 3954 2453 6407 1501 23.43 Orange to Riverside 3918 2295 6213 I623 26.12 Los Angeles to Kern 3201 2282 5483 919 16.76 Orange to San Bernardino 2386 2501 4887 115 2.35 San Bernardino to San Diego 2151 1317 3468 834 24.05 Riverside to San Diego 1781 1423 3204 358 11.17 aRefer to notes given in Table 9 above. Vn ro I j i FIGURE 5 INTERCOUNTY MIGRATORY PATHS EXCEEDING A GROSS VOLUME OF 3000 LICENSES ANNUALLY: SOUTHERN CALIFORNIA 1970 - 1973 IEOERB 0= O R O S S LICENSES ■=RET LICENSES =STREAN DIRECTION RIRTR IROPORTtORAL TO RET IRA 0=1411 ■-111 UR BERNARDINO 0-1041 R=1H1 6=12117 ■=iai 0=1100 ■ -11U Q=U04 LOS AROELES ■=U4 6=22111 N=1M2 Vj\ V j J Table 11, Migratory Paths Exceeding a Gross Volume of 3,000 Licenses Per Year which Connect a Northern County with a Southern County, California, 1970-1973 Paths Stream8 , Counter- Stream® Gross0 Netd Efficiency® Los Angeles to Santa Clara 3210 3313 8523 1897 22.26 Los Angeles to Alameda 4374 3170 7544 1204 15.96 Los Angeles to San Francisco 3193 258O 5773 613 10.62 Los Angeles to Sacramento 2792 1923 4715 869 18.43 Los Angeles to San Mateo 2189 1763 3952 426 10.78 Los Angeles to Contra Costa 2312 1467 3779 845 22.36 Los Angeles to Fresno 1850 1367 3217 483 15.01 aThe number of licenses from the first county surrendered in the second county. ^The number of licenses from the second county surrendered in the first county. c The sum of the stream and the counterstream. dThe difference between the stream and the counterstream. The net divided by the gross multiplied by 100. 55 FIGURE 6 INTERCOUNTY MIGRATORY PATHS EXCEEDING A GROSS VOLUME OF 3000 LICENSES ANNUALLY CONNECTING NORTHERN AND SOUTHERN COUNTIES: CALIFORNIA, 1970 - 1973 U N FRANCISCO O CONTRA COSTA 0=4111 0=411 0=1111 0=441 0=1144 0=1104 o = ms ■ = (is LEOEND O^OROSS LICENSES N=NET LICENSES t=STREAN DIRECTION RltTN PROPORTIONAL TO NET *■ 1 56 year is almost twice the sum of the net effects of all of the major northern streams (18,813). Much of the movement in the north is from San Francisco to bordering counties, while in the south the movements are from Los Angeles to bordering counties. The movement between northern and southern counties occurs exclusively as a result of flows from Los Angeles County in the south to, with the exception of Fresno, the bay area counties in the north (see Table 11 and Figure 6). These streams are netting the northern part of the state 6,337 licenses annually. The major migratory flows taking place within the state of California are occurring within and between the metropolitan areas of the state. The flows described above, including the out of state exchanges, suggest a pattern of successive population redistribution. The streams from out of state tend to be toward larger metro politan areas in the state. The streams from the larger metropolitan areas are toward surrounding metropolitan counties and in the case of Los Angeles County toward the northern metropolitan region. The exchanges experienced by the smaller metropolitan areas with out of state jurisdictions suggest streams of returning migrants. In the following chapter, the analysis of the intrastate intercounty migration patterns is undertaken. The research design employed prohibits the analysis of 57 out of state exchanges\ i.e.i the inability to establish values for the independent variables for out of state streams required that the analysis be restricted to a closed system of intercounty moves within California. The analysis focuses on the determinants of migration and the relative ability of each to explain the selection of a county of destination by intercounty movers. CHAPTER IV THE DETERMINANTS OF MIGRATION* AN EMPIRICAL ANALYSIS Empirical Indicators Independent Variables The assumption underlying this investigation implies that migration is something other than a random process. It is contended that prospective migrants select from among alternative destinations on the basis of a set of specific destination characteristics (determinants). Those identified and chosen for study here include popula tion size, the number of previous migrants, income level, employment, industrial and commercial development, housing availability and distance. These determinants have been quantified as follows* Population (P.)* the estimated January 1, 19?2 J (mid-period for the July 1, 1970 to July 1, 1973 migration data) population of each county j. (Source* California Statistical Abstract* 1972, State of California, Documents Section.) 58 59 Migrant Stock (MS.)i the number of people in each J county j, five years old or older who on April 1, 1970 reported having an April 1, 1965 place of residence in a different county in the state of California. (Source* U.S. Department of Commerce, Bureau of the Census, General Social and Economic Characteristics * California, Series PC (1)-C, 1970.) Income (1^)* ‘ the yearly average per capita labor income in each county j for the period January 1, 1970 to January 1, 1972. (Source* California County Fact Book* 1972, County Supervisors Associ ation of California, Sacramento, California.) Employment (E.)* the percent of the civilian labor J force employed in county j as of April 1, 1970. (Source* U.S. Department of Commerce, Bureau of the Census, General Social and Economic Charac teristics* California, Series PC (1)-C, 1970.) Industrial and Commercial Development (V.)* the J average annual reported valuation of non-residen- tial permits in county j for the period January 1, 1969 to January 1, 1973. (Source* Monthly Report of Building Activity in Cities and Counties of California, Research Department, Security Pacific National Bank, Los Angeles, California.) 60 Housing (H.)i the average annual number of housing J units authorized in county j for the period January 1, 1970 to January 1, 1973. (Source* Monthly Report of Building Activity in Cities and Counties of California, Research Department, Security Pacific National Bank, Los Angeles, California,) Distance (D. .)* the shortest highway mileage between A J the largest city of county i and the largest city of county j, (Source* Automobile Association of America, Maps and Highway Mileage Charts of California.) Dependent Variable This investigation is concerned with explaining the intercounty migration patterns within the state of California. A symptomatic indicator of the magnitude of migratory flows has been developed from the California Department of Motor Vehicles* change of address file des cribed at the beginning of Chapter III. From this file three summary matrices were constructed, one for each of three years beginning July 1, 1970 and ending July 1, 1973. Each matrix contains the number of licenses surrendered in any county j from any county i for a one year period. From these three matrices, a fourth summary matrix was computed representing the yearly average exchanges. The latter 61 matrix was employed as the representation of the magnitude of migration between any two counties under the assumption that the averaging process would yield greater stability. The DMV file is new and appears subject to fluctuations extraneous to actual fluctuations in migratory patterns.'*' As such, migration has been defined as follows* Migration ( ^)* the yearly average number of changes of address registered with the DMV for the three year period July 1, 19?0 to July 1, 1973, whose previous address was in county i and whose current address is in county j, (Source* California Department of Motor Vehicles.) Method of Analysis The system of migratory flows under analysis has been artificially closed and includes only the migration which taices place between the fifty-eight counties of California, i.e., out of state and APO streams have been excluded due to the inability to establish values for the independent variables under consideration. The approach chosen has been used by Gallaway, Gilbert and Smith in 196?» Greenwood in 1970 and again by Greenwood and Sweetland in 1972. Each ^part from the technical problems involved in developing a new file, the DMV procedure of including a change of address card in license plate renewal notices for only one of the three year periods appeared to cause fluctuations in reported changes. It might be noted that the 197^ renewals did not contain a change of address card. 62 county is taken as a point of origin from which migrants may flow to any of the other fifty-seven counties in the state, i.e., the fifty-seven streams from each of the fifty-eight counties, a total of 3,306 streams, are analyzed in fifty-eight different sets of multiple regres sions. In each case the variables in the analysis have been subjected to logarithmic transformations and thus logarithmic multiple regressions have been computed. The form of the equations thus becomes (with all symbols referring to the log of the variables previously defined)* Mij = a* Dij’b» Pj8» MSjf’ Ij6> Ejh» Vjn* Hjm» where‘ D. . = distancei P. * population! MS. = migrant stock; X J j J I. * income; E. = employment; V. = commercial and indus- J J J trial development; and = housing. In the above equation a is a constant (intercept) to be determined and b, e, f, g, h, n, and m are migration elasticities associated with each of the independent variables. It should be noted that attention will be focused upon the standardized beta weights rather than the estimated elasticities. Estimates of the independent con tribution made by each independent variable toward the explanation of the variation in migration (M. .) will be ^ J attempted in order to evaluate the relative importance of each as a determinant of migratory behavior. 63 Intercorrelation Among Independent Variables For each of the fifty-eight regressions, the inter correlation among six of the seven independent variables remains relatively constant. Working with a closed system of fifty-eight counties (N), consideration of each of the possible N-l destinations from each N possible origin fails to change significantly the intercorrelation among those variables relating to the characteristics of the areas of destination. The average intercorrelation among these six variables has been computed by summing across the fifty-eight intercorrelation matrices, the results of which are reported in Table 12. The seventh independent variable, distance, takes on a set of unique intercorrelations with the other six independent variables for each of the fifty-eight regres sions. Each of these fifty-eight sets is reported in Table 13. As shown in Table 12, extensive intercorelation exists among population size, migrant stock, commercial and industrial development and housing availability. Such results suggest that multicollinearity will be high and thus that interpretations of the standardized beta weights as an indication of the relative independent contribution of variables to the explanation of migration will have to be approached with extreme caution. 64- Table 12, Zero Order Logarithmic Correlation Coefficients Among Independent Variables3- P. 0 Ij vi H3 MS. J 1.00 .4^ .50 .95 .94 .99 Eo 1.00 • 34 .48 .46 .44 1,00 .53 .43 .50 vj 1.00 .92 .95 Hi 1.00 .94 MS. J 1,00 d t The variables represented include population (P^)f t l employment )„ income commercial and industrial development (V.), housing (H.) and migrant stock (MS.). J J « # While such intercorrelations suggest complications when using multiple regression techniques, they also serve to verify some of the implications discussed in Chapter II. It appears, for example, that the extent to which counties differ in size of population (P.) is in large part due to J the extent to which each county has been impacted by previous migration (MS^). Migrant stock explains ninety- eight percent of the variance in population size. With, population size serving as an indication of the extent of previous growth experienced by each county, it is also apparent that growth takes place where growth has taken 65 Table 13. logarithmic Correlation Coefficients between Distance and Each of the Other Independent Variables by County3- County Pj vs H3 MS . J Alameda -.159 -.187 -.315 -.204 -.180 -.162 Alpine .394 .134 .108 .279 .207 .370 Amador .195 .108 .049 .095 .089 .184 Butte .407 .305 .153 .415 .395 .410 Calaveras .180 .033 .043 .111 ,062 .168 Colusa .278 .288 .115 .286 .269 .269 Contra Costa -.060 -.124 -.233 -.116 -.103 -.065 Del Norte .341 .361 .149 .409 ,426 .356 El Dorado .456 .253 .008 .302 .357 .435 Fresno -.148 -.181 .058 -.172 -.199 -.134 Glenn .293 .306 .161 .313 .30 5 .394 .290 Humboldt .343 .423 .151 .404 .342 Imperial -.534 -.431 -.195 -.574 -.552 -.556 Inyo .151 -.133 -.050 -.038 -.027 .138 Kern -.468 -.396 -.084 -.485 -.494 -.463 Kings -.295 -.249 .025 -.325 -.339 -.265 Lake .159 .056 -.015 .162 .159 .160 Lassen . 568 .456 .193 .552 .564 .571 Los Angeles -.520 -.424 -.138 -.543 -.533 -.538 Madera -.187 -.142 -.029 -.204 -.244 -.171 Marin -.115 -.143 -.233 -.150 -.140 -.117 Mariposa .010 -.057 .088 -.077 -.105 .010 Mendocino .159 .085 .080 .169 .148 .149 Merced -.079 -.102 -.020 -.114 -.155 -.060 Modoc .535 .598 .197 .557 .571 .535 Mono .321 .109 .112 .161 .222 .299 Monterey -.191 -.176 -.139 -.230 -.234 -.185 Napa -.020 -.040 -.161 -.050 -.057 -.029 Nevada .324 . 220 .071 .285 .260 .315 Orange -.547 -.401 -.176 -.566 -.540 -.549 Placer .286 .203 .067 .234 .209 .273 Plumas .497 .436 .149 .479 .504 .494 Riverside -.507 -.387 -.214 -.526 -.499 -.512 Sacramento .194 .089 .033 .133 .108 .178 San Benito -.320 -.264 -.257 -.335 -.348 -.323 San Bernardino -.458 -.414 -.146 -.502 -.489 -.466 San Diego -.508 -.414 -.166 -.528 -.508 -.509 San Francisco -.173 -.203 -.295 -.192 -.214 -.180 San Joaquin .040 -.025 -.049 -.035 -. 066 .030 San Luis Obispo -.480 -.399 -.196 -.471 -.489 -.481 San Mateo -.196 -.171 -.285 -.238 -.223 -.196 Santa Barbara -.584 -.458 -.190 -.598 -.588 -.595 Santa Clara -.127 -.148 -.210 -.176 -.175 -.123 Santa Cruz -.207 -.191 -.281 -.252 -.242 -.207 66 Table 13 — Continued County p3 Ej VD Hj MS . J Shasta .425 .460 .238 .500 .469 .431 Sierra .446 .361 .167 .378 .367 .432 Siskiyou .452 .511 .245 .499 .500 .455 Solano -.015 -.111 -.108 -.057 -,06? -.014 Sonoma -.056 -.138 -.202 -.077 -.083 -.065 Stanislaus -.001 -.064 -.045 -.058 -.096 .000 Sutter .255 .215 .000 .234 .234 .236 Tehama .365 .448 .219 .417 .394 .369 Trinity .390 .516 .195 .427 .450 .399 Tulare -.2?7 -.228 -.019 -.290 -.314 -.268 Tuolumne .232 .025 .060 .111 .089 .223 Ventura -.567 -.428 -.250 -.567 -.554 -.576 Yolo .139 .070 .004 .113 .088 .134 Yuba .259 .166 .197 .242 .212 .259 The variables represented include population (P-)» employment (E.), income (I.), commercial and industrial t J u development (V.), housing (H.) and migrant stock (MS.). J J 3 place* i.e., as population increases, growth activities also increase. The size of population of the counties explains ninety percent of the variance in the yearly average valuation of non-residential permits and eighty-eight percent of the variance in the yearly average number of housing units authorized. These findings are consistent with expectations. Due to the problems involved when the magnitude of multicollinearity is high the relative importance of each of the independent variables to explain migration will be assessed from sets of multiple correlation coefficients. 67 McNemar’s (1969) suggestion that "(t)he relative importance of the N-l variables can be judged by the 0's or by ascertaining the reduction in R that results from dropping out, in turn, each predictor” (p, 213) will be employed. In the following analysis, both 0's (standardized regression coefficients which when squared are interpreted as estimates of the independent contribution of an indepen dent variable to the explanation of a dependent variable) 2 and reductions in R will be evaluated. Independent Variables and Migration The logarithmic correlation coefficient between each independent variable and the dependent variable of migra tion was computed for each of the fifty-eight counties. 2 Table 1^ contains the resulting r ' s, as well as the sign, in parentheses, of the computed correlation coefficient (r). The size of the population at destination (counties j) was consistently positively related to the magnitude of migration from each origin (county i). The proportion of variance in migration explained by population ranged from a low of .005 Tor Sierra County to a high of .905 for San Diego County. Of the fifty-eight coefficients, only that which was computed for Sierra County failed to be significantly different from zero at least at the .05 level. Only three of the remaining coefficients failed to 68 be significant at the .001 level. The average amount of variance explained by population over the fifty-eight counties was 59.9 percent. The migrant stock variable was also consistently positively related to the magnitude of migratory streams flowing from county i to other counties j. The proportion of variance explained across counties ranged from a low of .006 for Shasta County to a high of .920 for San Diego County, Once again only a single coefficient, in this case that computed for Shasta County, failed to be signi ficantly different from zero at least at the .05 level and all others with the exception of three were significant at the .001 level. Migrant stock was able to explain an average of 60.0 percent of the variance in migration across the fifty-eight counties. The distance which exists between county i and county j correlated negatively with the magnitude of migra tion in all but three counties. Of the fifty-eight computed correlation coefficients, however, twelve failed to be significantly different from zero at the .05 level. The proportion of variance explained ranged from a low of -.000 for El Dorado County to a high of ,538 for Imperial County. The average amount of variance explained over the fifty-eight counties was 19.7 percent. Both indicators of growth were found to be consis tently positively correlated with migration. In each case 69 (V. and H.) only two of the computed correlation coeffi- J J cients failed to he statistically significant from zero at the .05 level. These counties were Modoc and Sierra. The range of proportion of variance explained by yearly average valuation of non-residential permits was from a low of .006 to a high of .827. The average amount of variance explained was 55.0 percent. Residential permits accounted for only 1,5 percent of the variation in migra tion from Sierra County but was found to explain a high of 86,2 percent in the migration from Orange County. The average amount of variance explained by authorized units over the fifty-eight counties was 55*3 percent. Of the seven independent variables employed in this study* those serving as indicators of economic determinants were found to be the weakest in their zero order relation ship with migration. While income was found to have a consistently positive correlation with migration and pro duced only two coefficients failing to be significantly different from zero at the .05 level, it was only able to explain an average of 13.8 percent of the variance. Employment, on the other hand, was positively correlated with migration in all but two of the counties and produced twenty-six correlation coefficients which failed to be significantly different from zero at the .05 level and was able to account for an average of only 8,3 percent of the variance in migration across the fifty-eight counties. 70 In summary, population, migrant stock, growth in commerce and industry, and housing development all are interpreted as having consistent and approximately equally strong attraction for migrants. Income and employment, while displaying a consistent pattern of attractiveness, have only a mild impact on migration in relation to the above variables. Distance, as was expected, was found to be a consistent and moderate deterrent to the development of migratory streams. For those who are familiar with California counties it will be readily apparent from inspection of the results contained in Table 14 that the ability of each of the variables to explain migration decreases with size of origin county. At the extreme, those coefficients failing to be significant, for example, consistently are found in the smaller, and in many cases border counties. The fact that the system of counties has been artificially closed to include only movements within the state could be dis torting the results of the border county computations. On the other hand, it should be noted, flows of zero magnitude were not eliminated from the analysis and as a result, small counties such as Alpine where all but ten of the flows were zero, would also be subject to computational distortions. Apart from the distorting effects of a closed system and zero streams which are readily apparent among the small Table 14. Squared Logarithmic Correlation Coefficients8- between Each Independent Variable and Migration (M. .), by County, California, 1970-1973° J County P5 D. . 13 EJ V3 Hj MS. J Alameda .762 (“).207 .085 .170 .686 .756 .780 Alpine .152 .011* .015* .108 .142 .103 .122 Amador .322 (-).123 .016* .090 .375 .362 .315 Butte .564 (0.037* .030* .126 .437 ,442 • 555 Calaveras .282 (0.246 .033* .077 .333 .344 .289 Colusa .263 ( O .202 .018* .097 .230 .207 .246 Contra Costa .735 (0.156 .079 .167 .670 .713 .750 Del Norte .367 (0 .050* .010* .141 .327 .282 .371 El Dorado .665 .000* .110 .187 .665 .715 .694 Fresno .803 (0.249 .190 .124 .725 .756 .785 Glenn .302 (0.151 .011* .046* .265 .223 .283 Humboldt .599 (0.042* .015* .081 .464 .462 .607 Imperial .744 (0.538 .147 .192 .709 .657 .734 Inyo .438 (0.038* .116 .105 .481 .487 .429 Kern .849 (0.411 .188 .143 .76 9 .786 .841 Kings .754 (0.347 .155 .173 .719 .673 .726 Lake .625 (-).082 .127 .197 .607 .56 9 .630 Lassen .418 (-).OOl* .000* .141 .371 .337 .406 Los Angeles .857 (0.448 .192 .168 .793 .840 .889 Madera .636 (0.307 .093 .104 .570 .594 .616 Marin .691 ( O .206 .081 .219 .668 .670 .729 Mariposa .400 (O.220 .067* .089 .418 .371 .361 Mendocino .557 (0.107 .033* .104 .520 .458 .561 Table 14 — Continued County D. • 13 E . 3 *3 V3 H3 MS. Merced .715 (-).233 .133 .142 .660 .697 .696 Modoc .087 (-),121 (-).039* .012* .065* .041* .0 75 Mono .242 .009* .088 .069 .248 .209 .248 Monterey .831 (-).166 .087 .148 .735 .772 .828 Napa .698 (-M54 .054* .154 .640 .657 .719 Nevada .506 (-).071 .035* .186 .482 .503 .530 Orange .881 (-).434 .165 .171 .793 .862 .901 Placer .607 (-).054* .039* .140 .556 .598 .626 Plumas .236 (-).102 .000* .141 .190 .167 .237 Riverside .867 (-).477 .142 .174 .778 .825 .884 Sacramento .711 (-).051* .062* .130 .617 .680 .728 San Benito .599 (-).369 .134 .150 .586 .553 .579 San Bernardino .869 (-).429 .168 .175 .777 .819 .882 San Diego .905 (-).353 .176 .239 .809 .859 .920 San Francisco .759 (-).259 .149 .259 .696 .745 .783 San Joaquin .749 (-).098 .078 .109 .727 .721 .7^3 San Luis Obispo .878 (-),416 .161 .194 .827 .807 .878 San Mateo .754 (-).248 .104 .191 .695 .725 .77 4 Santa Barbara .888 (-).506 .170 .212 .825 .843 .908 Santa Clara .810 (-).173 .086 .161 .744 .782 .823 Santa Cruz .801 (-),180 .070 .190 .723 .730 .801 Shasta .494 (-).057* .066* .071 .360 .361 .491 Sierra .005* (-).279 - .065* .002* .006* .015* .006* Siskiyou .450 (-).030* .000* .099 .389 .365 .434 Table 14 — Continued County ?0 D. . 10 E. 0 Vj H o MS. J Solano .764 (-),114 .098 .170 .715 .703 .780 Sonoma .687 (-).167 .065* .163 .584 .591 .704 Stanislaus .735 (-).137 .098 .105 .670 .711 .731 Sutter .531 (-).125 .053* .152 .486 .437 .537 Tehama .339 (-),163 .000* .051 .264 .237 .326 Trinity .148 (-).163 .036* .062 .110 .071 .139 Tulare .795 (-).295 .136 ,132 .709 .697 .77 6 Tuolumne .499 (-0.093 .055* .135 .548 .536 .489 Ventura .871 (-).514 .155 .205 .7 88 .830 .88 5 Yolo .725 (-).076 .101 .156 .659 .634 .719 Yuba .543 (-).no .045* .112 .489 .483 .530 Average .599 .197 .083 • V-O CO .550 .553 .600 aThe signs in parentheses represent the direction attached to the unsquared coefficients (r), v o Those r 's which are not significant at least at the ,05 level are signified by an asterisk (*). 74 border counties, there appears to be a general tendency for size of origin to be related to the ability of the independent variables to explain migration. This observa tion is noted here with the intention of anticipating the need to control for county size in an effort to fully elaborate on the operation of the model under considera tion. Migrant Stock and Population Migrant stock has been empirically defined as the number of people (five years old or older) residing in each county on April 1, 1970 who were reported as having an April 1, 1965 place of residence located in a different county within the state. While similar census definitions have been used by others concerned with the effect of previous migration on present or future migration patterns, such a definition fails to be totally compatible with the frame of reference concerning migrant stock developed earlier. When evaluating the differential selection of points of destination by migrants from a single place of origin, migrant stock would "ideally" include those residing at each possible point of destination who had previously resided at the place of origin under analysis. Due to the lack of availability of information which would serve as an appropriate approximation of this conceptual "ideal," conclusions concerning the role of migrant stock 75 as a determinant of further migration appears to have been rendered inappropriate in the present analysis. The first indication that a serious problem existed was the extremely high intercorrelation which was found between migrant stock and the size of population (r = .99). As should have been expected, large populations are large as a result of previous magnitudes of in-migrants. As such, the two variables would appear to be tautological. Second, the correspondence between the ability of migrant stock and the ability of population size to independently explain migration was also almost perfect. The rank order correlation coefficient of the r 's between population and 2 migration and the r * s between migrant stock and migration over the fifty-eight counties was .9991. In evaluating the relative contributions of migrant stock and size of population in a multiple regression model of migration, an effort was made to reduce the high inter correlations which existed among other independent variables. To this end, the number of residential units authorized was eliminated from the regression equation, and the valuation of industrial and commercial permits was divided by the size of population producing a per capita value (FV ^), The alterations improved substantially the state of the independent variable intercorrelation matrix (see Table 15). 76 Table 15. Intercorrelations (logarithmic) among Independent Variables, Reduced Model P3 MS . J e j h w j p3 1.00 .99 .44 .50 .15 MS. J 1.00 .44 .50 .16 Ej 1.00 .34 .27 1.00 .24 ^0 1.00 aThe variables indicated include population (P.= ), J migrant stock (MS.), employment (E.), income (I*) J J J and per capita valuation of commercial and industrial development (PV.), J While the potential problems which might have been caused by the growth factors were eliminated by employing the above set of procedures, the high intercorrelation between migrant stock and population itself will produce extensive multicollinearity. By using a set of regressions, however, dependence upon beta weights for interpretation is reduced. The equations to be evaluated include the followingi Model A. "ii ■ Dir Er Model B. "ij =Eir Er Ir pj Model c. "is = Dir Ei’ V MS. Model D. Mij - V V pj * »s3 - 77 o The squared multiple correlation coefficients (R ' s) resulting from these equations, as well as the zero order 2 r *s between migrant stock and migration and population and migration, are reported in Table 16. Acting alone, migrant stock and population have virtually equal ability to explain migration. For the fifty-eight counties, population was able to explain on the average 59.9 percent of the variance in migration while migrant stock was able to explain an average of 60.0 percent of the variation. Working in conjunction with the distance, employment, income and per capita valuation variables, which by them selves were able to explain only an average of 33.7 percent of the variance (Model A), population and migrant stock once again rendered similar results. Population (Model B) was able to increase the variance explained from 33.7 percent to 79.4 percent, an increase of 45.7 percent, while migrant stock (Model C) increased the percent explained to 79.1 percent, an increase of 45.4 percent above the 33.7 percent explained by the other four variables in the model. When both population and migrant stock are allowed to appear in the same model (D), the independent contri bution which each makes is minimal. The independent 2 2 contribution made by migrant stock (R ^ - R is less than one percent (.803 - .794 * .009 or .9 percent) while o Table 16, Proportion of Variance Explained (R ) in Migration from County i to County j by Population at j (P^)» Migrant Stock at j (MS.) and Models A thru J D for Each County i, J California, 1970-1973 Models County MS. J Aa B* C° Dd Alameda .762 .780 .305 .894 .908 .908 Alpine .152 .122 .124 .176 .153 .247 Amador .321 .315 .252 .585 .566 .588 Butte .564 .555 .219 .880 .871 .882 Calaveras .282 .289 .366 .670 . 666 .671 Colusa .263 .246 .362 .647 .618 .663 Contra Costa .734 .750 .283 .881 .891 .891 Del Norte .367 .371 .229 .603 .617 .617 El Dorado .666 .694 .241 .835 .851 .851 Fresno .803 .785 .450 .945 .936 .945 Glenn .303 .283 .255 .646 .613 . 660 Humboldt .599 .607 .166 .8 77 .883 .886 Imperial . 745 .734 .633 .859 .839 .865 Inyo .438 .429 .202 .537 .519 .538 Kern .84-8 .841 .529 .909 .904 .910 Kings .755 .7 26 .545 .881 .874 .881 Lake .626 .630 .339 .812 .817 .819 Lassen .417 .406 .160 .686 .670 .686 Los Angeles .857 .889 .55 8 .908 .927 .929 Madera .637 .616 .422 .816 .806 .816 Marin .691 .729 .362 .846 .880 .888 Mariposa .401 .361 .352 .633 .591 .674 Mendocino .557 .561 .247 .802 .795 .803 Merced .7X6 .696 .411 .894 .887 .894 Modoc ,088 .075 .157 .471 .437 .485 Mono .241 .248 .139 .254 .259 .259 Monterey .830 .828 .294 .916 .914 .919 Napa .699 .719 .277 .880 .893 .894 Nevada .507 .530 .290 .795 .815 .815 Orange .882 .901 .535 .913 .929 .929 Placer .607 .626 .223 .853 .863 .864 Plumas .235 .237 .289 .66 3 .663 • 666 Riverside .867 .884 .560 .941 .952 .952 Sacramento .711 .728 .213 .896 .902 .905 San Benito .599 .579 .451 .752 .731 .757 San Bernardino .869 .882 .551 .941 .948 .9% San Diego .904 .920 .521 .924 .937 .938 San Francisco .759 .783 .435 .894 .912 .913 San Joaquin .750 .743 .232 .916 .896 .916 San Luis Obispo .878 .878 .525 .941 .938 .942 San Mateo .753 .774 .372 .878 .897 .897 79 Table 16 — Continued Models County P5 MS . U A B C D Santa Barbara .887 .908 .616 .933 .945 .945 Santa Clara .810 .823 .29 6 .930 .943 .944 Santa Cruz .801 .801 .297 .897 .893 .899 Shasta .494 .491 .189 . 861 .861 .866 Sierra .005 .006 .311 .436 .431 .436 Siskiyou .450 .434 .169 .766 .742 .767 Solano .764 .780 .284 .892 .906 .906 Sonoma .68? .704 .289 .859 .869 .870 Stanislaus .734 .731 .271 .904 .896 .905 Sutter .530 .537 .313 .847 .835 .847 Tehama .340 .326 .275 .792 .77 4 .792 Trinity .147 .139 .282 .540 .532 .540 Tulare .796 .766 .440 .898 .885 .899 Tuolumne .498 .489 .273 .758 .735 .760 Ventura .8?0 .885 .595 .929 .936 .936 Yolo .724 .719 .280 .892 .880 .892 Yuba .5^3 .530 .295 .846 .828 ,846 Averages .599 .600 .337 .794 .791 0 C D • X j = Dij» V V PTr X j = Dij’ V X* pr Xj = DiJ* V V X* "sr Xj = Dij’ Er X X- MS., P J 80 2 2 that made by population (R d - R c) is only slightly above one percent (.803 - .791 = ,012 or 1,2 percent). As a result of these findings and due to the lack of confidence expressed earlier in the ability of the adopted empirical definition of migrant stock to adequately measure the concept as it has been theoretically employed, further consideration of this variable appears unwarranted at this time. While these arguments are felt to apply to such a variable when disaggregated migration streams are being evaluated, it should be noted that a different conclusion would be appropriate for studies concerned with migration at the aggregated level. Simply, the above set of findings appears to confirm Greenwood's (1970) and Nelson's (1959) contention that population will "pick up the effects of the migrant stock variable if the migrant stock variable is not itself included. . ." (Greenwood, 1970*383). Model B With the elimination of the migrant stock variable now supported and retaining the alternations previously mentioned, i.e., the elimination of the residential permit variable and the conversion of the valuations variable into a per capita figure, a more comprehensive evaluation of Model B (where migrant stock is excluded as a deter minant) would appear to be warranted. Such is undertaken prior to the reintroduction of the growth indicators in an 81 effort to evaluate a set of determinants having a notice ably reduced potential for excessive multicollinearity. Table 17 contains the 0 's (squared standardized beta coefficients) for each of the variables in Model B for each of the fifty-eight counties under analysis. The 2 resulting squared multiple regression coefficients (R 1 s) have also been reported. The signs in parentheses refer to the direction attached to the unsquared betas. The level of significance of each of the coefficients has also been indicated. 2 Squared standardized beta coefficients (B 1 s) are interpretated as the amount of variance explained in the dependent variable by an independent variable with all other independent variables held constant. The B *s for each determinant have been averaged across the fifty-eight counties to provide a summary estimate of each of their relative contributions. The results indicate conclusive support for two of the basic hypotheses under study. First, the most important determinant of migration proved to be population. The average squared standardized beta over the fifty-eight counties was .753. Not only was this the strongest average coefficient among the variables in the equation but each of the population coefficients was in the predicted direction, with all but three proving to be statistically different from zero at the .001 level Table 1?. Logarithmic Regression Coefficients8- and Coefficients of Multiple Determination of Migration*3 (M. .), by County, California, 1970-1973° — J County D. • 13 Ei Ij R2 Alameda .873*** 1 — ,134*** (— .025** (-).012 .000 ,894*** Alpine .120 (- .002 (- .005 .029 (-).ooo .176 Amador .524*** (- .185*** (- .039 (-).ooo .027 .585*** Butte 1.158*** (- ,338*** (- .009 (-).005 (-).004 .880*** Calaveras .473*** (- .348*** (- .018 (-).OOl .022 .670*** Colusa ,474*** (- .401*** (- .003 .002 .001 .647*** Contra Costa .902*** (— ].142*** (- .022** (-).Oll .000 ,881*** Del Norte .604*** (- .246*** (- .014 .004 .019 .603*** El Dorado 1,211*** (” ,244*** .000 (-).015 (-).OOl .835*** Fresno .784*** (- .132*** (- .000 (-).007* .000 ,945*** Glenn .625*** (— .355*** (- .006 (-).006 .006 ,646*** Humboldt 1.142*** (- .261*** {- .010 (-),022* .000 .877*** Imperial .435*** (- .174*** (- .013 .004 .000 .859*** Inyo .568*** (- .077* (- .002 (-).006 .007 .537** Kern .706*** (- .068*** (- .000 (-).004 .000 .909*** Kings .563*** (- .145*** (- .004 .004 .001 .881*** Lake .745*** (- .183*** (- .001 (-).ooo .014 .812*** Lassen 1.046*** (- .288*** (- .039* .002 .003 ,686*** Los Angeles .647*** (- .067*** (- .001 (-).OOO .002 .908*** Madera .620*** (- .178*** (- .007 (-).004 (-).ooo .816*** Marin .741*** (- .137*** (- .029* (-).OOl .009 .846*** Mariposa .443*** (- .224*** (- .008 .000 .001 .633*** Mendocino .861*** (- .212*** (- .042** (-).005 .016 .802*** 00 PO Table 17 — Continued County pi D. . 13 Ej Ij pvj R2 Merced .743*** (—) .176*** (- .002 (- .003 .000 .894*** Modoc .576*** (-) .437*** (- .023 ( - .009 .006 .471** Mono .217** (-) 1001 .004 (- .000 .006 .254 Monterey .95^*** (-).068*** (- .024** (- .008 (-).ooo .916*** Napa .927*** (-) ,155*** (- .039** (- .011 .003 .880*** Nevada .839*** (-).288*** (- .016 .001 .003 ,795*** Orange .732*** I M J ,o44*** (- .001 (- .001 .000 ,913*** Placer 1.025*** (-),238*** (- .020 (- .005 .001 .853*** Plumas .692*** ( - ) ,496*** (- .007 .008 (-).002 .663*** Riverside .707*** (-) .097*** (- .009* (- .003 .000 .941*** Sacramento 1.069*** (“) ,174*** (- .018* (- .008 (-).OOl .896*** San Benito .470*** (-)!l64*** (- ,002 (- .005 .006 ,752*** San Bernardino .690*** (-) ,099*** V .005 (- .000 (-).OOl .941*** San Diego .774*** ( -) .028*** (- .002 .002 (-).OOO .924*** San Francisco .683*** (-) .139*** (- .004 .000 .000 .894*** San Joaquin 1.050*** (-) .119*** (- .025** (- .030*** .009* .916*** San Luis Obispo .741*** (-).073*** (- .010* (- .002 .008* .941*** San Mateo .764*** (-).130*** ( - .010 (- .006 .001 .878*** Santa Barbara .654*** ( ") ,071*** (- .007 .000 .001 .933*** Santa Clara .940*** (-),108*** (- .024*** (- .010* .002 .930*** Santa Cruz .921*** (-).078*** ( - ,039*** (- .005 .000 .897*** Shasta 1.129*** (-) ,400*** ( - .009 (- .010 .000 .861*** Sierra .232** (-) ,438*** ( .043 (- .002 (-).ooo .436** Siskiyou .995*** (”) .325*** t l- .027 (- .001 .017 .766*** 00 V j J Table 1? — Continued County po D. . Ej PV. 2 R2 Solano .930*** (-).115*** ( - ).021** (-).006 .004 ,892*** Sonoma .884*** (-).163*** (-).032** (-).009 (-).OOl .859*** Stanislaus .982*** (-).145*** (-).013* (-),028** .000 .904*** Sutter .862*** (-).337*** (-).002 (-).005 .001 ,84?*** Tehama .818*** (-).473*** ( -).0 1 3 {-).004 .005 .792*** Trinity 05*** (-).325*** (-).0 3 8 .011 (-).ooo .540** Tulare .756*** (-),102*** (-).004 (-).005 (-).ooo .898*** Tuolumne .781*** (-),222 (-),028* (-).ooo .012 .758*** Ventura .651*** (-).089*** (-).008* (-).ooo .000 .929*** Yolo .95^*** (-).164*** (-).005 (-).0 0 7 .001 .892*** Yuba .870*** (-).306*** (-).013 (-).ooo .001 .846*** Averages .753 .195 .015 .006 .004 .79^ degression coefficients are squared standardized betas. The sign in parentheses refers to the direction attached to the unsquared coefficients, ^Independent variables include population (P.), distance (D. .), employment (E.)f income (I -), and per capita valuations of ^ commercial ana*1 industrial ^ development (PV.). J °Levels of significance of coefficients as indicated by asterisks are* *p<.05, **p<.01f and ***p<.001. 00 - p - 85 of significance. Of the three which failed to reach this level, two were found to he significant at the .01 level with only one, Alpine County, failing to produce an F value sufficient to establish it as different from zero with ,05 being the minimum acceptable level of signifi cance. These results strongly confirm the contention that migrants tend to move toward populated areas, all else being equal. Second, distance is shown to have a decidedly restrictive effect upon migration. With an average 2 0 = .195* it appears as the second most important variable in the model. For all fifty-eight counties the direction of the computed coefficient is negative, i.e., as distance between counties increases migration decreases. Once again all but three of the coefficients were found to be statis tically different from zero at the .001 level of signifi cance. Of these three, one was found to be significant at the .05 level with the remaining two failing to be different from zero (Alpine, £ = .002 and Modoc, £ = .001). It is thus concluded that distance does in fact act as an impediment to migration. The remaining three variables in the equation failed to make a significant contribution to the explanation of migration above and beyond that made by population and distance. For all three the average £ 's across the fifty-eight counties were approaching zero (i.e., F = .015, .006 and .004 for employment, income and per capita valuations respectively). For both the employment and income variables a large proportion of the coefficients were in a direction opposite to what had been hypothesized (96,6 percent 56 and 77.6 percent 45 respectively). Once again, as others have found (Greenwood, 1970; Greenwood and Sweetland, 1972; and Tarver, 1965) while both income and employment when considered by themselves appear to be making a contribution to the explanation of migration, once they are considered in conjunction with the population variable their contribution is significantly decreased and in many cases the direction of their relationship with migration reversed. It appears that the effects and influence of previous migration, which might have been toward high income and high employment areas, overrides the current influence of these variables on present migratory patterns. Growth and Migration To this point it is apparent that migration is toward populated areas and is attenuated by distance. It also appears that the current economic state of an area of destination exerts no "pull" on prospective migrants. Returning now to the original hypothesized model, excluding the migrant stock variable, consideration will be given to the impact of the growth indicators as determinants. The equation being considered is of the formi 87 M. . = P., D. ., E -, I., V-, H., where* P. = population* ^•3 3 J J J J D = distance* E. = employment* I. = income* i J J J V. = commercial and industrial development and J H. = housing. The squared standardized beta coefficients J 2 (3 ) and the squared multiple correlation coefficients (R ) which resulted are reported in Table 18 for each county. Once again, population and distance dominated as the principle determinants of intercounty migration. Popula tion was the strongest variable in the model with an average beta square of .632. Distance appeared as the _2 second strongest variable with a 3 = .193. For both variables all coefficients were in the hypothesized direction. The growth variables were found to be the next _2 strongest determinants in the model. For housing the 3 was equal to .057* while for commercial and industrial _2 development the 3 equalled . 04l. The direction of the coefficients, however, was only sporadically in the pre- _2 dieted direction. The 3 for the economic variables was once again found to be approaching zero (E. = .015 and J I. = .007) with the direction of their coefficients being J fairly consistently in the wrong direction (opposite to that predicted). While the inclusion of the growth indicators _2 (V. and H.) appeared to only slightly reduce the 3 for J J Table 18. Logarithmic Regression Coefficients8 , and Coefficients of Multiple Determination of Migration*5 (M. .), by County, California, 1970-1973° ^ J County P3 D. . ij e j V. 0 Hj R Alameda .36*** -.12*** -.03*** -.00 -.00 .14** .908*** Alpine .44 -.01 -.00 .02 .00 -.23 .195 Amador .01 -.17*** -.04 .00 .19 .06 .590** Butte 2.18 -.34*** -.01 -.01 -.01 -.11* ,890*** Calaveras .00 -.32*** -.02 -.00 .22 .05 .678*** Colusa .56* -.41*** -.00 .00 .03 -.05 .652*** Contra Costa .55*** -.13*** -.03** -.01 .00 .04 .884*** Del Norte .09 -.26*** -.02 .00 .18 .01 ,607*** El Dorado .65** -.20*** -.00 -.01 -.01 .15* .847*** Fresno .77*** -.13*** -.00 -.01 .00 -.00 .945*** Glenn .69* -.35*** -.00 -.01 .10 -.12 .660*** Humboldt 1.41*** -.25*** -.01 -.03** .01 -.05 .882*** Imperial .56*** -.18*** -.01 .00 .00 -.02 .861*** Inyo .05 -.05 -.00 -.00 .08 .08 .543** Kern .65*** -.07*** -.00 -.00 .00 .00 ,909*** Kings .62*** -.15*** -.00 .00 .02 -.03 .885*** Lake .23* -.19*** -.00 -.00 .14 .00 .815*** Lassen .89** -.29*** -.04* .00 .04 -.01 .689*** Los Angeles .22** -.06*** -.00 .00 .00 .11* .918*** Madera .73*** -.18*** -.01 -.01 .00 -.01 .817*** Marin .17 -.13*** -.03** .00 .06 .06 .852*** Mariposa .78* -.26*** -.01 -.00 .04 -.18 .650*** Mendocino .64** -.22*** -.03* -.01 .22* -.09 .814*** 00 CD Table 18 — Continued County Pj D. . 10 vo Ho R2 Merced .64*** -.17*** -.00 -.00 .00 .00 .894*** Modoc .65* -.42*** -.02 -.02 .10 -.12 .484** Mono .51 -.01 .00 -.01 .05 -.17 .264 Monterey .80*** -.07*** -.03** -.01 .00 .01 .916*** Napa .56*** -.15*** -,04*** -.01 .03 .01 .881*** Nevada .36* -.28*** -.02 .00 .02 .04 .800*** Orange .28*** -.05*** -.00 .00 -.00 .15** .929*** Placer .72*** -.23*** -.02* -.00 .00 .01 .854*** Plumas 1.02** -.49*** -.01 .01 -.01 -.01 .663*** Riverside .41*** -.10*** -.01** -.00 -.00 .07* .948*** Sacramento 1.23*** -.17*** -.02** -.01 -.02 .00 .897*** San Benito .30* -.17*** -.00 -.01 .08 -.02 .755*** San Bernardino # 66*** -.10*** -.01 .00 -.02 .02 .943*** San Diego .41*** -.03*** -.00 .01 -.01 .10** ,934*** San Francisco .35** -.13*** -.01 .00 .00 .06 .900*** San Joaquin .?4*** -.13*** -.02** -.04*** .10* -.01 .919*** San Luis Obispo .37*** -.07*** -.01* -.00 .07* .00 .941*** San Mateo .43** -.12*** -.01 -.00 .00 .04 .882*** Santa Barbara .28*** -.0?*** -.01* .00 .00 .05* .938*** Santa Clara .51*** -.10*** -.03*** -.01 .01 .03 .934*** Santa Cruz .77*** -.08*** -.04*** -.01 .01 .00 .897*** Shasta 1.31*** -.39*** -.01 -.02 .03 -.05 .867*** Sierra .07 -.41*** -.05 -.00 -.01 .09 .444* Siskiyou .31* -.34*** -.03 -.00 .14 .01 .769*** 00 VO Table 18 — Continued County pd Dij Eo vj Hi R Solano .70*** -.12*** -.02* -.01 .04 -.00 .893*** Sonoma 1.36*** -.17*** -.03** -.01 -.00 -.04 .862*** Stanislaus .97*** -.15*** -.01* -.03** .01 -.00 .904*** Sutter 1.18*** -.35*** -.00 -.01 .04 -.12* ,860*** Tehama .89*** -.47*** -.01 -. 01 .09 -.10 .840*** Trinity .78* -.30*** -.03 .01 .00 -.10 .550** Tulare 1.09*** -.11*** -.00 -.01 .00 -.06 .904*** Tuolumne .30 -.22*** -.03* -.00 .10 .00 .758*** Ventura .29*** -.09*** -.01* .00 -.00 .08* .938*** Yolo 1.24*** -.18*** -.00 -.01* .03 -.09* ,901*** Yuba .93*** -.31*** -.01 -.00 .02 -.03 ,849*** Averages .632 .193 .015 .007 .041 .057 .800 Regression coefficients are squared standardized betas. The sign in parentheses refers to the direction attached to the unsquared coefficients. v Independent variables include population (P.)t distance (D..), employment (E.)» income (I.)» commercial and industrial ^ development ^ (V.), and * * housing * * (H.). J J cLevels of significance of coefficients as indicated by asterisks are* *P<.05, **p<.01f and ***p<,001. \o o 91 population from .753 (Table 1?) to .632, the actual impact appears underestimated as a result of the inflationary effect of multicollinearity on the population betas (i.e., nine of the fifty-eight estimated betas for popula tion exceeded 1,0). It should also be noticed that the consistency in the significance of the computed coeffic ients for population which was reported earlier has suffered substantially with the inclusion of the growth indicators into the model. In an effort to evaluate the independent contri butions made by each of the sets of variables! i.e., popu lation, distance, economics and growth, the selected set of regressions reported in Table 19 was computed for each county. Also contained in Table 19 is the average amount of variance explained by each model (regression) over the fifty-eight counties. The two variables, population and distance, by themselves have the ability to explain an average of 77.6 percent of the variance in migration for the fifty-eight California counties. Adding the economic and growth variables increases the average amount of variance explained by only 2.4 percent. Note, however, that without population the average amount of variance explained is only 4.1 percent (independent contribution) below the average 80 percent explained when all of the variables are included 92 in the model. The independent contribution of the growth factors on the other hand was 1.0 percent. _2 Table 19. Average Correlation Coefficients (R ) across Counties for Selected Models of Migration, California, 1970-1973a Models R A. “ 13 s P3 .599 B. “ij S Dij .197 C. “ij = p3’ d13 .776 D. “13 =r P3’ D13* e 3* *3 .790 E. “ 13 s p3' d13' e3* *3' v3’ h3 .800 F. “ 13 rs d13’ e 3’ I 3 .332 G • “ 13 r= d13’ e 3’ V V h3 .759 Variables indicated include migration (M^), population (Pj), distance (Dij). employment (E^) income (Ij)i commercial and industrial development (V^) and housing (Hj). v Figures reported are the average squared correlation coefficients, both zero order and multiple, across the fifty-eight counties in California, 93 An interesting effect to note is the erosion of the independent contribution made by the population variable with the successive inclusion of each of the other major indicators. By itself population explains 59.9 percent of the variance in migration. When distance is introduced the independent contribution of population remains vir tually unchanged (i.e., R 2pd - R 2d = .776 - .197 = .579). With the inclusion of the economic variables, population's independent contribution is only mildly reduced to 45.8 percent (i.e., R 2pdei - R 2dei = .790 - .332 = .458). Finally, by including the growth factors into the model, the independent contribution made by population to the explanation of migration is reduced to only 4.1 percent RZpdeivh * RZdeivh = -800 * *759 » .041). Model G The above analysis strongly suggests that the effects of population size can be disaggregated by inspecting the growth dimensions independent of the population variable. In an effort to evaluate the extent to which such is possible, the coefficients resulting from Model G (where population has been excluded as a determinant) are reported in Table 20. These results, unlike those pre viously reported, are presented with the origin counties appearing according to the size of population. This latter form of presentation is intended to facilitate the Table 20. Logarithmic Regression Coefficients8- and Coefficients of Multiple Determination of Migration*5 (M. .), by County, California, 1970-1973° j County Du E5 V0 H0 R Los Angeles -.254*** -.068 .053 .235* .580*** .902*** Orange -.223*** -.083 ,039 .179 .668*** .909*** San Diego -.176** -.089 .117* .200 .650*** .906*** Alameda -.326*** -.200*** -.027 .242 .696*** .884*** Santa Clara -.280*** -.190*** -.035 .409*** .566*** .900*** San Francisco -.336*** -.094 .070 .258 .569*** .877*** San Bernardino -.290*** -.105 .041 .248 .570*** .898*** Sacramento -.324*** -.185** -.105 .353* .632*** .820*** Contra Costa -.324*** -.186** -.034 .350* .592*** .849*** San Mateo -.324*** -.132* .011 .333 .541*** .853*** Riverside -.322*** -.135** .011 .230* .596*** ,920*** Fresno -.340*** -.034 -.039 .417** .457*** ,894*** Ventura -.316*** -.130** .049 .232* .564*** .918*** Kern -.259*** -.055 -.019 .388** .440*** .866*** San Joaquin -.277*** -.184** -.137* .686*** .353** .875*** Santa Barbara -.272*** -.120* .072 .303** .508*** .919*** Monterey -.221*** -.193** -.035 .412** .558*** .863*** Marin -.347*** -.195** .027 .421** .467** .841*** Sonoma -.376*** -.211** -.047 .482** .420** .772*** Stanislaus -.310*** -.148* -.119 .512*** .470** .845*** Tulare -.313*** -.092 -.039 .536*** .310* .832*** Solano -.302*** -.171** -.034 .576*** .390** ,847*** Santa Cruz -.237*** -.231*** .018 .472** .484*** .847*** v O ■ P - Table 20 — Continued DiJ E3 Vi Hi R2 San Luis Obispo -.279*** -.125* -.009 .542*** .331** .917*** Merced -.375*** -.077 -.016 .407** .451** .853*** Butte -.557*** -.137 -.012 .560** .447* ,746*** Humboldt -.568*** -.106 -.119 .653*** .411* .794*** Yolo -.368*** -.097 -.049 .652*** .302* .821*** Napa -.359*** -.229*** -.052 .493** ,473*** ,845*** Shasta -.696*** -.112 -.066 .705*** .369* .784*** Placer -.411*** -.192** -.006 .421* .571*** .811*** Imperial -.406*** -.133 .081 .414* .239 .824*** Kings -.365*** -.087 .080 .511** .240 .844*** Mendocino -.462*** -.217* -.062 ,827*** .122 ,771*** Yuba -.523*** -.143 .017 .572** • 3^5* .788*** El Dorado -.314*** -.072 -.012 .263 ,758*** .821*** Sutter -.549*** -.172 -.045 .685*** .220 .783*** Madera -.399*** -.122 -.029 .419* .357* .769*** Siskiyou -.608*** -.173 .006 .635** .409* .749*** Tehama -.715*** -.123 -.044 .731*** .181 .745*** Nevada -.483*** -.171* .086 .397* .519** .778*** Tuolumne -.393*** -.188* .014 .547** .352* .743*** Lake -.428*** -.050 .012 .585*** .310* ,800*** San Benito -.398*** -.054 -.053 .532** .170 .735*** Glenn -.597*** -.097 -.068 .701** .095 .614*** Lassen -.498*** -.231* .056 .612** .385 .631*** Inyo -.181 -.063 -.034 .378 .396 .541** vO Table 20 — Continued County D. . 13 Ei V3 Ho R Del Norte -.535*** -.126 .077 .558* .276 .601*** Calaveras -.561*** -.149 -.022 .498* .248 .678*** Colusa -.627*** -.082 .068 .487* .197 .615*** Amador -.400*** -.212* .013 .477* .298 .590*** Plumas -.664*** -.136 .145 .35^ .422 .596*** Trinity -.598*** -.180 .110 .497 .118 .500** Modoc -.654*** -.169 -.086 .677* .084 .441** Mariposa -.423*** -.105 .009 .614* .053 .603*** Mono .015 .061 -.020 .477 -.002 .251 Sierra -.609*** -.249* .015 .027 .428 .440** Alpine -.010 -.093 .167 .623 -.306 .179 Average 0 .178 .022 .004 p-\ - d - C V l • .186 .759 Regression coefficients are unsquared standardized betas. Independent variables include distance (D..), employment (E.)f income (I.), commercial and industrial development (V.) and housing (H-)* ^ J J °Levels of significance of coefficients as indicated by asterisks are* *P<.05* **p<.01l and ***p<.001. VO On 97 assessment of the effect of size of origin county on the ability of the model to predict migration. The strongest single variable in the model, once population is removed, was industrial and commercial _2 development (0 « .2^3). Each of the coefficients was in the predicted direction with a marked tendency for only those of middle size counties to be statistically signifi cant from zero. Of the fifty-eight coefficients, thirteen failed to be significant at the .05 level. The second strongest variable was housing development _2 with a 3 = .186. Unlike industrial and commercial development, the significance of the computed coefficients declined systematically with size of county. Twenty of the fifty-eight coefficients failed to be significantly different from zero at the .05 level. For two counties. Mono and Alpine, the coefficients were found to be in a direction opposite to that hypothesized. The third strongest variable to appear in the model _2 was distance, having a 0 = .178, only slightly less than that found for housing. The distance coefficients were consistently significantly different from zero and in the predicted direction. The only major exception was that of Mono County where the coefficient was neither significant nor in the predicted direction. Virtually all of the coefficients for employment were in a direction opposite to that predicted with less than 98 half being significantly different from zero at the .05 level. The coefficients for income were sporadically in the opposite direction with only two attaining significance at the .05 level. In both cases, income and employment, _2 the @ were approaching zero (.022 and .004 respectively). Conclusions Population and distance have been found to be strong determinants of migration. It has also been shown that most of the effects of population can be disaggregated into at least the two dimensions of industrial and commercial development and housing availability. It is also apparent that these effects articulate systematically with size of county (Table 20). In Chapter V differences in the ability of the determinants to explain migration across county types will be examined more fully. CHAPTER V COUNTY DIFFERENTIALS Introduction To this point the analysis has described the relative importance of each of a set of determinants to the explana tion of intercounty migration within California. The determinants considered were destinational characteristics. Origin characteristics have been held constant by analyzing the movements from each of the counties individually. The findings (especially those reported in Table 20) suggest that the model and its components woric differently for different origin counties. To further evaluate the extent to which such is the case the counties have been categor ized according to a typology based on migratory activity. While the magnitude of migratory activity being experienced by a county correlates strongly with county size, the use of activity patterns was found to be con ceptually more consistent with an attempt to explain county differentials in the influence of the migration determinants. The categories which have been developed 99 100 follow the logic of Alonzo and Medrich's (1970) classifi cation system which concentrated on Spontaneous Growth Centers (SGC's). Spontaneous Growth Centers have been defined as areas which are "growing without the benefit of special assistance" (Alonso and Medrich, 1970* 2). The SGC classification was applied to all Standard Metropolitan Statistical Areas (SMSA's) which were experiencing growth rates greater than double the national average (p. 3). SMSA's in close physical proximity to SGC's and which were found to be growing as a result of SGC expansion were defined by Alonso and Medrich as Novea (p. 27). County Types An evaluation of the annual magnitude of gross intra state migration (GIM) experienced by each county between July, 1970 and July, 1973 suggested four distinct cate gories of county types (see Table 21). Among the fifty-eight California counties, twelve were found to be experiencing GIM's greater than 25,000 licenses annually. These twelve counties comprise the northern and southern Metropolitan Growth Centers (MGC's) and account for 81.9 percent of all population growth occurring in California since April 1, 1920. As counties of origin this set accounts for 75.^ percent of all movements to other counties within the state. 101 Table 21. Intrastate Migration Activity, by County, within County Types, California, 1970-1973 Annual Exchanges With Other Counties To Other From Other Gross County Counties Counties Net (GIM) METROPOLITAN GROWTH CENTERS Los Angeles 144,430 71,900 - 72,530 216,330 Orange 41,730 66,560 + 24,830 108,290 Santa Clara 30,410 34,830 + 4,420 65,240 Alameda 33,440 29,840 - 3,600 63,280 San Diego 24,260 35,970 + 11,710 60,230 San Francisco 30,510 18,760 - 11.750 49,270 San Mateo 25,640 22,160 — 3,480 47,800 San Bernardino 22,400 24,610 + 2,210 47,010 Contra Costa 19,120 20,730 + 1,610 39,850 Sacramento 19,050 20,470 + 1,420 39,520 Riverside 15,800 22,110 + 6,310 37.910 Ventura 11,950 17.150 + 5,200 29,100 NOVEA AREAS Fresno 10,380 Santa Barbara 9,640 Kern 9.530 Marin 9,080 Sonoma 6,370 San Joaquin 8,200 Monterey 7,3^0 Santa Cruz 4,820 Solano 6,450 Stanislaus 5,540 10,910 + 530 21,290 10,420 + 780 20,060 9.350 - 180 18,880 9,030 - , 50 18,110 10,940 + ^.570 17,310 7.530 mm 670 15,730 7,850 + 510 15.190 8,690 + 3,870 13.510 6,200 mm 250 12,650 6,590 + 1,050 12,130 TERTIARY AREAS San Luis Obispo 4,160 7,010 ♦ 2,850 11.170 Butte 4,270 6,100 1.830 10,370 Tulare 4,860 5.^90 + 630 10,350 Yolo 4,700 5.550 ♦ 850 10,250 Placer 3.490 5,410 1,920 8,900 Humboldt 3.110 4,180 + 1,070 7.290 Merced 3,500 3.580 + 80 7,080 Shasta 2,910 4,000 + 1,090 6,910 Napa 2,800 3,790 + 990 6,590 El Dorado 2,270 4,200 + 1,930 6,470 102 Table 21 — Continued Annual Exchanges With Other Counties County To Other From Other Counties Counties Gross Net (GIM) RURAL AREAS Mendocino 1,910 2,750 + 840 4,660 Imperial 2,410 2,180 mm 230 4,590 Kings 2,285 2,085 - 200 4,370 Yuba 2,170 2,000 - 170 4,170 Sutter 2,060 2,030 30 4,090 Madera 1,570 1,760 + 190 3,330 Tuolumne 820 2,360 + 1,540 3,180 Nevada 1,140 1,910 + 77 0 3,050 Lake 800 1,870 + 1,070 2,670 Siskiyou 920 1,630 + 710 2,550 Tehama 1,130 1,350 220 2,480 Lassen 640 960 + 320 1,600 Inyo 590 980 + 390 1,570 Calaveras 560 970 410 1,530 Plumas 500 860 + 360 1,360 Glenn 650 690 + 40 1,340 Amador 440 8?0 + 430 1,310 Del Norte 370 930 + 560 1,300 San Benito 630 610 w 20 1,240 Mono 280 690 + 410 970 Tulare 340 580 + 240 920 Colusa 420 470 + 50 890 Mariposa 280 600 + 320 880 Modoc 230 340 + 110 570 Sierra 100 140 + 40 240 Alpine 30 55 + 25 85 FIGURE 7 103 CALIFORNIA COUNTIES AS CLASSIFIED BY ANNUAL MAGNITUDE OF GROSS INTRASTATE MIGRATION (GIM): 1970 - 1973 Oregon LLLUJJ.J Ml ^Oiy 101+ The second set of counties identified, conforming to Alonso and Medrich*s Novea Areas (NA), were counties with GIM's between 12,000 and 25,000 licenses per year. With the exception of Fresno County this set of ten counties lies on the borders of MGC's, Serving as counties of origin, these areas account for 13.9 percent of all move ments to other counties. The MGC's and the NA's taken together account for 89.3 percent of all such movements. A third set of counties identified, those having GIM's between 6,000 and 12,000 annually, include the three northern isolated urban counties of Humboldt, Shasta and Butte, as well as seven other urban hinterlands surrounding the MGC's and NA's. These areas were found to account for 6.5 percent of all movements to other counties. This group is identified as Tertiary Areas (TA). The remaining twenty-six counties, those with GIM's of less than 6,000 licenses annually comprise the category of Rural Areas (RA). These counties are the smallest (population wise) and generally those farthest removed geo graphically from the MGC's. Together this group accounts for only *+.2 percent of movements to other counties. Data relating to the evaluation of the determinants of migration has been summarized by county type in Tables 22, 23 and 2k. It is apparent from these syntheses that the impact of the determinants articulates systemati cally by county type. 105 Population. Distance and Migration Zero Order Correlations It has been consistently found that migration is toward populated areas and is inhibited by distance (Galloway, et. al,, 196?* Greenwood, 1969aj Lowry, 1966* Miller, 1972 and Zipf, 19^6). From the results reported in Table 22, however, it is apparent that the extent to which these generalizations hold is markedly dependent upon the type of county from which the migration is occurring. Population at destination, by itself, was able to explain an average of 59.9 percent of the variance in the migration flows from the fifty-eight counties in California. Examination by county type, however, produced noticeable differences in the impact of population upon migration patterns (see Table 22), Among the twelve MGC's and the ten Novea counties the impact of population was noticeably higher than the statewide average. Population at destination explained 81.5 percent of the variance in the flows from MGC's and ?8.0 percent of the variance in the flows from NA's. Among the Tertiary Areas the influence of population remained above the statewide average yet fell noticeably short of the impact noted in the MGC's and the NA's (67.3 percent of the variance in migration explained). The impact of population on the Table 22. Summary Statistics of the Interrelationships between the Variables Population, Distance and Migration, by County Type, California, 1970-1973 County Type (GIM) No. Of Counties M. . Out 2a l j p Percent m.pd _2 2 ^ 2 added - mp by p mp.d JL 2 ^ 2 - added md by d md.p 2 R joint cont. MGC 25,000 - 225,000 12 75.4 .898 .815 .586 .852 .312 .083 .423 + .229 NA 10,000 - 25,000 10 13.9 .877 .780 .654 .843 .223 .097 .434 + .126 TA 6,000 - 12,000 10 6.5 .865 .673 .729 .844 .136 .192 .558 -.056 RA Below 6,000 26 4.2 .647 .401 .491 .584 .156 .246 .425 -.090 MGC's & NA's 22 89.3 .888 .799 .617 .848 .272 .089 .428 + .182 MGC's, NA's & TA's 32 95.8 .881 .760 .652 .847 .230 .121 .469 + .108 Total 58 100.0 .776 .599 .580 .729 .197 .177 .449 + .019 Subscripts refer to the variables migration (m), population (p), and distance (d). ^Joint contribution = 2gJ,r„,, p d pd H o o\ 10? flows from Rural Areas was half that recorded from the Metropolitan Growth Centers (40.1 percent of the variance in migration explained). County type differentials in the effect of distance upon migration were also evident. Both among the Metro politan Growth Centers and the Novea Areas the effect of distance was stronger than the statewide average of 19.9 percent of the variance in migration explained (31.2 per cent and 22.3 percent respectively). The influence of distance on the migration from the Tertiary and Rural Areas was about half that recorded for the Metropolitan Growth Centers (13.6 percent and 15.6 percent respectively). While these outcomes suggest a pattern of differen tial impacts, the actual interrelationships between population, distance and migration appear comfounded. Population and distance, taken together, explained an average of 77.6 percent of the variance in the migration taking place from the fifty-eight counties in the state. The major differences by county type were between the rural and non-rural counties. For the non-rural counties (i.e., the MGC's, the NA's and the TA's) the ability of population and distance to explain migration averaged over eighty-eight percent (89.8 percent, 87,7 percent and 86,5 percent respectively. For the rural counties, however, population and distance explained only 64.7 percent of the variance in migration. 108 The Multicollinearity Factor Efforts to assess the relative contribution of population and distance to the explanation of migration revealed the degree of complexity which exists in the interrelationships of these three variables. Among the MGC's and the NA's, the average squared multiple correla- _2 tion coefficient (R ) was smaller than the sum of the _2 _2 average zero order values (r + r md)» while among the 2 2 2 TA's and the RA's the R was greater than r mp .+ r md (see Table 22 for an explanation of notation). These differences are produced by the differences in the multi collinearity factor of the multiple regression model <R2 = d2p + S2d + 29pPdrpd), where 29p9drpd equals the magnitude of multicollinearity. In the case where 0p is positive and 0d is negative, the sign of the multicol linearity factor is dependent upon the direction of the relationship between population and distance (rpd). For the MGC's and the NA's, the relationship between population and distance is generally negative, i.e., being physically part of the growth centers, large populations are close. In this case the multicollinearity factor (joint contribution) will be positive* i.e., (2)(0^)(-0d)(-rdp) would equal a positive multicollinear ity value. On the other hand, among the TA's and RA's the relationship between population and distance tends to be direct (positive). In this case the multicollinearity 109 factor will be negative t i.e., (2) (+0p) (-0^) (tr^) would equal a negative multicollinearity value. Distance, when taken by itself, appeared to have the largest impact upon migration from the Metropolitan Growth Centers and decreased in importance across county type. When distance is considered in conjunction with population, its independent contribution to the explanation of migration follows a reverse pattern. The independent contribution made by distance is greatest among Rural _2 Areas (R added « .246) and least among Metropolitan _2 Growth Centers (R added - .083). Population, on the other hand, when considered alone was found to have an impact among the MGC's twice that found for the Rural Areas. In the multiple model, however, the independent contribution made by population to the explanation of migration from the Metropolitan Growth _2 Centers (R added « .586) was only 9.5 percent more than its independent contribution among the Rural Areas _2 (R added « .491). In fact, using independent contribution as a criteria, population was most important to the _2 _2 Tertiary (R added *= .729) and Novea (R added = .654) Areas. Suppression Effects With the direction of the relationship between distance and population changing somewhat systematically 110 across county types and due to the fact that population is directly correlated with migration while distance is inversely correlated with migration, it appears likely that the independent variables are involved in suppressive relationships with each other. In an effort to reflect upon this conclusion the partial correlation coefficients rmp d and rmd p were computed. It should be noted that in all cases, controlling for one of the independent variables increased (over the zero value) the relationship between the other independent variable and migration. These outcomes confirm the existence of suppression effects and lead to a further set of interpretations concerning the relative impact of population and distance upon the explanation of migration. Once the effect of distance was controlled for, population was found to be of equal importance in the explanation of migration from the urban counties. Among this group of thirty-two counties, the amount of variance in migration explained by population, once distance was controlled for, ranged from a low of 84,3 percent for the Novea Areas to a high of 85.2 percent for the Metropolitan Growth Centers, The influence of population upon migration from Rural counties (RA's), however, was found to be noticeably lower than that recorded for the urban counties (58.4 percent of the variance explained). Ill Distance, on the other hand, once population had been controlled for, was found to be most important to the migrants from the Tertiary Areas (55.8 percent of the variance explained) and of lesser and yet virtually equal importance across the remaining categories. The amount of variance in migration explained by distance, once popula tion was controlled for, among these latter areas ranged from 42.3 percent for the Metropolitan Growth Centers to 43.4 percent for the Novea Areas. The figure for the Rural Areas was 42.5 percent. Recapitulation Each of the above discussions lends perspective to the attempted understanding of the interrelationships which exist between population, distance and migration. Two of the findings reported above deserve repeating. First, among the thirty-two urban counties, which account for 95.8 percent of all movements under analysis, less than 16 percent of the variance in migration is left unexplained by population when the effects of distance are controlled. Second, taicen together, population and distance were able to explain an average of 88,1 percent of the variance in migration from the same set of thirty-two counties, leaving only slightly less than 12 percent of the variance unexplained. Both of these findings strongly suggest that the original Zipf (1946) 112 hypothesis is far from "being in danger of being discarded. Such findings strongly suggest that a closer look be given to the Zipf mechanism and formulation concerning patterns of migration. Determinants of Migration by Countv Type Model Effects While it has been reported that the two variables of population and distance by themselves do a good job of _2 explaining migration (R . = .776), the intent has been TO i pQ to evaluate a broader set of determinants. Adding to these two basic determinants the factors of income, employ ment, industrial and commercial development and housing, increased the percent of variance in migration explained by an average of only 2.4 percent (Table 23). This figure, _2 the amount of R added by the above mentioned four variables, was consistent across county types. The values ranged from a low of 2,0 percent added for the MGC's to a high of 2.7 percent added among the Novea Areas. Removing population from the six variable model reduced the proportion of variance explained by an average of only 4.1 percent. The greatest impact was among the _2 Tertiary Areas where the R dropped by 6.4 percent. The least impacts occurred in the extreme groups where the loss was 3.2 percent and 3.5 percent respectively for the MGC's and the Rural Areas. _2 Table 23. Average Multiple Correlation Coefficients (R ) for the Five Variable and Six Variable Models* by County Type, California, 1970-1973 County Type _2& Rra.pdievh 2 R added by ievh _2 Rm.dievh 2 R added by p Metropolitan Growth Centers .918 .020 .886 .032 Novea Areas .904 .027 .857 .047 Tertiary Areas .886 .021 .822 .064 Rural Areas .673 .026 .638 .035 Metro & Novea .911 .023 .873 CO 0 • Metro, Novea & Tertiary .903 .022 00 • .046 Totals .800 .024 .759 .041 Subscripts refer to the variables migration (m), population (p), distance (d), income (i), employment (e), commercial and industrial develop ment (v), and housing (h). 114 _2 The overall R 's for both the five variable and the six variable models were distributed similar to the _2 R ,'s, i.e., varying directly with type of county. In in i pci all cases the major point of differentiation was between the three urban categories and the rural counties. For the thirty-two urban counties, which together account for 95.8 percent of all movements to other counties, the six variable model was able to explain 90,3 percent of the variance in migration and the five variable model 85.7 percent of the variance in migration. For the rural counties these figures were 67.3 percent and 63.8 percent respectively. Component Effects The contention that population effects can be dis aggregated into dimensions has been supported in Chapter IV. In an effort to further articulate the impact of the distance, economic and growth determinants once population has been removed from the model. Table 24 has been prepared. In Table 24, the average zero order squared _2 correlation coefficients (r * s) of each of the five deter minants with migration are reported by county type. These _2 are accompanied by the 0 's resulting from the five variable multiple regressions. The first point to notice is the striking similarity _2 between the £ ' s for distance as found in Table 24 and the Table 24, Average Squared Logarithmic Zero Order Correlation Coefficients and Squared Multiple Regression Coefficients for Each Independent Variable with Migration (M..), by County Type, California, 1970-1973 ^ J Distance Employment Income Develonment Housing County Type _2 r md 2 0 d _2 r me 2 * e _2 r mi 2 0 • K 1 _2 r mv 2 P V 2 r mh 2 » h Metropolitan Growth Centers -.312 -.087 .130 -.020 .184 .004 .737 .079 .786 .364 Novea Areas -.223 1 • O CO VO .113 -.028 .158 vr\ 0 0 « 1 .714 .228 .728 .210 Tertiary Areas -.136 -.178 .085 -.018 .138 1 • 0 0 VjJ .598 .290 .607 .213 Rural Areas -.156 -.254 .035 -.023 .096 V> 0 0 * 1 .332 .306 .298 .084 Metro & Novea -.272 -,088 .122 -.023 .172 .004 .727 .147 .760 .294 Metro, Novea & Tertiary -.230 -.116 .110 -.022 .162 -.004 • 686 .192 .712 .269 Total -.197 00 r —I • I .083 -.022 .138 -.004 .550 .243 .553 .186 A — “ Average squared standardized regression coefficients (0 ) resulting from multiple regression equation: M. . = D. E., I., V., H.. H H 116 _2 R *s added by distance as found in Table 22, These results suggest two conclusions. First, distance's con tribution to the explanation of migration is steady and uneffected by the inclusion of population into the model, _2 Second, that the 0 's can be used as an adequate indicator of the independent contribution being made by each of the determinants. Once again it is noted that the relative importance of distance varies inversely with county type, being most important for those migrating from Rural Areas and least important for those migrating from Metropolitan Growth Centers, The zero order effects of both employment and income vary directly with county type, being most important to the migrants from the MGC's and least important for those migrating from the Rural Areas, Once other variables are entered into the model, however, the independent effects of _2 both of these variables approach zero (0 = ,022 for _2 employment and 0 = , 004 for income). It is apparent that population, or as in this case the growth dimensions, pick up the effects of the economic variables. It has been demonstrated that by replacing popula tion with industrial and commercial development and housing availability the amount of variance in migration explained drops by only 4,1 percent. As becomes obvious in Table 24, the effects of these two variables across county types run opposite to one another. 11? While in both cases, that of industrial and commercial development and that of housing availability, the zero order relationships were found to vary directly with county type, in the larger model the independent con tributions each made ran counter to each other. The impact of commercial and industrial development varied inversely with county type while the impact of housing varied directly with type of county. Industrial and commercial development was of greater _2 importance to migrants from Rural Areas (0 *= .306) than to migrants from the Metropolitan Growth Centers 2 (f = .079). On the other hand, housing was most important _2 to migrants from Metropolitan Growth Centers (3 = .36*0 and least important to the migrants from the Rural Areas _2 (3 = .084). While these figures reflect the differences in the extremes, the Novea and Tertiary Areas conform to the pattern suggested. Noticeable differences in the impact of industrial and commercial development occurred between the MGC counties and the counties in the other three categories. For the housing dimension, the notice able breaking point was between the rural and non-rural counties. The relative importance of each of the determinants maintained a clear pattern by county type. Migrants from the Metropolitan Growth Centers tended to be attracted _2 first by housing (3 = .364), only moderately by industrial _2 and commercial development (9 = .079) and only moderately _2 constricted by distance (9 = ,087). The migrants from the Novea Areas were equally attracted by industrial and _2 _2 commercial development (9 = ,228) and housing (9 - .210) - 2 and only moderately deterred by distance (9 = .089). Those migrating from the Tertiary Areas tended to be more attracted by commercial and industrial development _2 _2 (9 - .290), then by housing (9 = .213) and obviously _2 conscious of distance (9 * .178). For those migrating from the Rural Areas, there was a strong thrust toward _2 industrial and commercial development (9 - .306), with distance playing a major role in the decision to migrate _2 (9 = .25*0 and housing given only mild consideration _2 (9 = .084). These characteristics would appear to typify the major patterns of migration unfolding in California. The first and strongest pattern consists of the process of suburbanization. Streams of suburbanization are formed by migrants overflowing from the MGC's into the NA and TA counties. These streams were found to be predominantly in the direction of housing development. The other deter minants in the model were given only mild consideration by those leaving the MGC's, The second pattern suggested by the analysis repre sents the process of urbanization. Most typical of this pattern are the streams which flow from the Rural Areas. 119 These streams were found to be directed primarily toward commercial and industrial development rather than housing. For those leaving Rural Areas distance served as an important criteria in the selection of a destination. Movements from the Novea and Tertiary Areas appeared to represent a mix of the processes of suburbanization and urbanization. The data suggests that these areas tend to typify areas of transition. Since the areas themselves represent a blend of the county types (some Tertiary Areas closely approximating rural counties and some Novea Areas bordering on MGC status), it is not surprising that the migration patterns exhibited approximately a blend of the migratory processes described. CHAPTER VI SUMMARY AND CONCLUSIONS It can be conclused from the results reported in Chapters IV and V that approaching the explanation of migration from a "pull force" perspective is in fact fruitful. Using reported changes of address from one California county to another by drivers license holders for the period July lt 1970 to July 1, 1973* an evaluation of the impact of destinational characteristics upon popu lation redistribution has been accomplished. By evaluating the movements from each county in California to each of the other California counties, on a county by county basis, it was found that an average of eighty percent of the variance in migration could be explained by six destina tional determinants. The determinants which were used in a multiple regression model included the distance between counties, the size of population at destination, the level of income and employment at destination, industrial and commercial development and housing availability. 120 121 Beyond the verification of the validity and utility of the "pull force" perspective, the findings also support Zipf's (19^6) original migration hypothesis. Migration, when all else is said, can he fairly accurately explained as a function of population and distance. These two variables alone explained an average of 77.6 percent of the variance in migration. Once population and distance were allowed to explain what they could, the other four variables were able to increase the proportion of variance in migration explained by an average of only .02^. Given that the variables of income, employment, industrial and commercial development and housing avail ability failed to contribute significantly to the explana tion of migration once population and distance were considered, one might ask, why the continued presence of these variables in the analysis? A third major finding of this investigation was that population, when considered in conjunction with a larger set of determinants, fails to make a significant independent contribution to the explana tion of migration. It was found, for example, that by dropping population from the six variable model the average proportion of variance explained fell by only .0^1. It is apparent that most of the population variable's con tribution to the explanation of migration is jointly shared by the growth indicators of industrial and commer cial development and housing availability. The elimination 122 of population from the model allowed for the disaggre gation of its effects on migration. With population removed* it became obvious that movement was most predom inantly toward industrial and commercial development _2 _2 (p = .24-3) and secondarily toward housing (p * .186). The results of this investigation were also supportive of the contention that population (or its sub stitute dimensions of industrial and commercial develop ment and housing) when included in the migration model picks up the effects of the economic variables of income and employment. While income explained an average of 13.8 percent of the variance in migration at the zero order level* when either population or its dimensions were included in the model its independent contribution was _2 reduced to zero (p = .004). The same effect occurred for employment which at the zero order level explained an average of 8.3 percent of the variance in migration and yet was unable to make an appreciable independent contri- _Z bution to the multiple model (p - .022), The majority of the contribution made by distance to the explanation of migration was independent of that made by other variables in the model. By itself* distance was able to explain an average of 19.7 percent of the variance in migration. In conjunction with population its indepen dent contribution fell by only 2 percent to 17.7 percent. In the five variable model (that which excluded population 123 from the regression) the independent contribution of distance was found to be virtually equal to its zero order _2 average variance explained (3 = .197). County Types The findings summarized to this point refer to generalizations which can be applied to all counties in the state. The analysis undertaken which considered the influence of county types upon these generalizations, however, form a second set of major findings resulting from this investigation. Evaluation of the determinants by county type revealed the extent to which migratory activity within the state is concentrated among the counties of the Metropoli tan Growth Center and Novea Area types. The twenty-two counties classified into these two groups account for almost 90 percent (89.3 percent) of all movements to other counties. Evaluation of zero order relationships between the determinants and migration revealed systematic variation of their impacts across county types. The influence of population was found strongest among Metropolitan Growth _2 Centers (r = .815) and weakest among Rural Areas _2 (r = .401), The impact of distance was also found to vary directly with county type. For MGC counties the average 2 -2 r was .312, while for the RA counties r = .156, At the 1 2 4 zero order level this general pattern was found to apply to each of the six determinants being studied. When considered in conjunction with population, the magnitude of variance added by distance varied inversely with county type. Among the MGC counties distance added 8.3 percent to the variance explained in migration by population. For the NA counties the figure was 9.7 percent. Among the TA counties the percent of variance explained added by distance jumped to 19.2 percent and continued to increase to 24.6 percent for counties of the RA type. These figures were found to conform closely to _2 the 0 *s for distance by county type resulting from the five variable regression. Another major finding resulting from the "by county type" analysis was the systematic articulation of the impact of industrial and commercial development and housing availability. Industrial and commercial development maintained an inverse relationship with county type. Among _2 the Metropolitan Growth Centers the 0 was only .079 while _2 among the Rural Areas the 9 = .306. Housing availability, on the other hand, was found to vary directly with county _2 type. Among the Metropolitan Growth Centers the 0 = .364 _2 while among the Rural Areas 0 = .084. The six variable model was found to explain an average of 91.8 percent of the variance in migration among the twelve MGC counties. This figure was 91,9 percent 125 when the ten NA counties were included and 90.3 percent for the thirty-two non-rural counties combined. Consider ing that the thirty-two non-rural counties make up 95*8 percent of all movements within the state, the ability to explain an average of 90,3 percent of the variance in migration for these counties surpasses expectations. It should be noted that without population in the model, this figure would be 85.7 percent and that with population and distance alone the figure would be 88.1 percent. General Patterns These findings support the speculation that two general patterns are evident in the migratory processes taking place within the state of California. First, there appears to be a process of suburbanization. This is typified by those migrants moving from the Metropolitan Growth Centers. The analysis confirmed that for these people the dominant tendency was to move toward areas where housing development was taking place and only secondarily toward commercial and industrial development. The second process that was apparent was that of urbanization. This was more typical of those migrants moving from rural counties. Their movements were found to be decidedly toward commercial and industrial development and barely influenced by housing. 126 While the Metropolitan Growth Centers and Rural Areas provide extreme cases of these processes, it was evident that the movements taking place from the Novea and the Tertiary Areas exhibited properties of both types of processes. Serving as areas of transition in the pattern of population redistribution the migrants from these counties tended to display an equal interest in housing and commercial and industrial development. Conclusions It is the nature of science to be an ongoing process. Definitive answers are difficult to come by. What has been demonstrated here, as summarized above, is felt to be important. Just as important, however, has been the degree of confirmation offered concerning the utility of the DMV change of address file as a source of migration data. With time, improvements and further analysis it is believed that this file will come to serve as a model for the nation. Efforts are currently being made to improve the character and accuracy of the change of address file. While the data available approximates the magnitudes of movements taking place, two major questions remain unanswered. The first concerns the tendency to report a change of address while the second concerns the amount of time lapse which exists between a move and the reporting of the move. 12? It is a fairly sound assumption that the tendency to report a change of address to the DMV is not constant across individuals or areas. The tendency to report a change of address more than likely is different for different parts of the population, as well as for differ ent geographical areas of the state. The length of time "between a move and the reporting of the move is also probably mitigated by the characteristics of the population moving and the areas to which they move. These consider ations reflect on the ability of the file to accurately represent the relative magnitudes of movements taking place. Another major consideration facing those who wish to use the file as a mechanism for estimating net migra tion (essential for population estimates) is the conver sion factor(s) for translating licenses into people. More than likely we will find that such factors also articulate fairly clearly with the characteristics of movers and the characteristics of the areas from which and to which they move. Each of these questions remains as a problem demanding further investigation. The questions posed by the problems of population redistribution become more crucial each year. The impact of limited resources is being felt today as never before in this country. To reiterate Morrison (1973) "the exer cise of the right to move" ... has already begun to 128 impinge upon., "the nature of the entire society" (p. 1). The need to know and quest for understanding concerning the patterns of population redistribution is no longer a luxury but rather a necessity. BIBLIOGRAPHY Abbott, Walter P. 1972 "The gravity and Opportunity Models of Migration* A Theoretical Synthesis." Paper prepared for pre sentation at the 67th Annual Meeting of the American Sociological Association (August), 28-31* Alonso, William 1971 "The System of Intermetropolitan Population Flows." Working paper No. 155* prepared for the National Commission on Population Growth and the American Future. Alonso, William and Elliott Medrich 1970 "Spontaneous Growth Centers in Twentieth Century American Urbanization." Working paper No. 113, unpublished work supported by a grant from the Economic Development Administration of the United States Department of Commerce. Anderson, T.R. 1955 "Intermetropolitan Migration* A Comparison of the Hypotheses of Zipf and Stouffer." American Socio logical Review 20 (June), 287-291. 1956 "Intermetropolitan Migration* A Correlation Analysis." American Sociological Review 61 (March), 459-462. Beals, R.E., M.B. Levy and L.N. Moses I967 "Rationality and Migration in Ghana." Review of Economics and Statistics 49* 480-486. Blalock, Hubert M. 1972 Social Statistics. New York* McGraw-Hill Book Co. Blanco, C, 1963 "The Determinants of Interstate Population Move ments." Journal of Regional Science 5* 77-84. 129 Bodenhofer, H.J, 196? "Mobility of Labor and the Theory of Human Capital." Journal of Human Resources 2 (Fall), 431-448. Bogue, Donald J. 1959 "Internal Migration and Residential Mobility." Pp. 375-^18 in The Population of the United States. Glencoe, Illinois* The Free Press. Bowles, S. 1970 "Migration as Investment* Empirical Tests of the Human Investment Approach to Geographic Mobility." Review of Economics and Statistics 52, 356-364. Brennan, M.J. 1965 "A More General Theory of Resource Migration." In Patterns of Market Behavior. Edited by M.J. Brennan. Providence* Brown University Press. Bright, Margaret L. and Dorothy S, Thomas 1941 "Interstate Migration and Intervening Opportunities American Sociological Review 6, 773-783. Cowgill, Donald 0. 1961 "Value Assumptions in Recent Research on Migration. Sociological Quarterly 2 (October), 281-292. Davis, Kingsley and Frederick G. Styles (eds.) 1971 California’s Twenty Million* Research Contributions to Population Policy, Population Monograph Series, No. 10. Institute of International Studies, University of California, Berkeley. Department of Commerce, U.S. Bureau of the Census 1963 Census of Population* i960 Subject Reports, "Mobility for States and State Economic Areas." Final Report PC (2) 2B. Washington, D.C.i U.S. Government Post Office. 1966 Current Population Reports. No. 339 (June 6), 25. 1967 County and City Data Book, 1967. A Statistical Abstract Supplement. Washington, D.C.* U.S. Government Post Office. United States Census of the Population* i960. Subject Reports, Lifetime and Recent Migration. Final Report PC (2) 2D. Eldridge, H.T. 1965 "Primary, Secondary, and Return Migration in the United States." Demography lit 444-455. 131 Fabricant, Ruth A. 1970 "An Expectational Model of Migration." Journal of Regional Science» Vol. 10, No. 1, 13-24-. Gallaway, L.E. 1967 "Industry Variations in Geographic Labor Mobility Patterns," Journal of Human Resources 2, 461-474. Gallaway, L.E., R.F. Gilbert and P.E. Smith 1967 "The Economics of Labor Mobility* An Empirical Analysis." Western Economic Journal. V»2 (June), 211-223. 1968 "Reply." Western Economic Journal. VI*3 (June), 244. Galle, O.R. and K.E, Taeuber 1966 "Metropolitan Migration and Intervening Opportuni ties," American Sociological Review 31> 5-13. Goldstein, Sidney 1955 "Migration and Occupational Mobility in Norristown, Pennsylvania," American Sociological Review 20 (August), 402-4oB"I 1958 Patterns of Mobility. 1910-1950. Philadelphia* University of Pennsylvania Press. 1964 "The Extent of Repeated Migration* An Analysis Based on the Danish Population Register." Journal of the American Statistical Association. LIX (December), 1121-1132. Greenwood, M.J, 1968 "The Economics of Labor Mobility an Empirical Analysis* Comment." Western Economic Journal. VI*3 (June), 243. 1969a "An Analysis of the Determinants of Geographic Labor Mobility in the United States." Review of Economics and Statistics 51 (May), 189-194. 1969b "The Determinants of Labor Migration in Egypt." Journal of Regional Science 9# 283-290. 1970 "Lagged Response in the Decision to Migrate." Journal of Regional Science 10, 375-384. 1971a "A Regression Analysis of Migration to Urban Areas of a Less Developed Country* The Case of India," Journal of Regional Science 11, 253-262, 1971b "An Analysis of the Determinants of Internal Labor Mobility in India," Annals of Regional Science 5, 137-151. Greenwood, M.J, and P.J, Gormaly 1971 "A Comparison of the Determinants of White and Nonwhite Interstate Migration," Demography 8, 141-155.' Greenwood, M.J. and Douglas Sweetland 1972 "The Determinants of Migration Between Standard Metropolitan Statistical Areas." Demography, Vol. 9, No. 4 (November), 665-681. Hagerstrand, T. 1957 "Migration and Area." In Migration in Swedent A Symposium. Lund Studies in Geography, Series B, No. 13, 27-158. Lund* C.W.K. Gleerup Publishers. Haenszel, William 1967 "Concept, Measurement, and Data in Migration Analy sis." Demography 4, 253-261. Hamilton, C.H, 1965 "Practical and Mathematical Considerations in the Formulation and Selection of Migration Rates." Demography 2, 429-443. Hamilton, Horace C. 1959 "Educational Selectivity of Net Migration from the South." Social Forces 38, 33-42. Hathaway, Dale E. 1967 "Occupational Mobility from the Farm Labor Force." In Farm Labor in the United States. Edited by C.E. Bishop, New York* Columbia University Press, Ikle, Fred C. 1955 "Comment on Theodore R. Anderson's Intermetropolitan Migration* A Comparison of the Hypotheses of Zipf and Stouffer." American Sociological Review 20 (December), 713-714. Isaac, J, 1947 Economics of Migration. Oxford* Oxford University Press, Isbell, E.C, 1944 "Internal Migration in Sweden and Intervening Oppor tunities." American Sociological Review 9, 627-639. 133 Jackson, J.A. 1969 "Migration - Editorial Introduction." In Migration. Edited by J.A, Jackson. Cambridge* Cambridge University Press, 1-10. Jansen, C.J, 1969 "Some Sociological Aspects of Migration." In Migration. Edited by J.A, Jackson. Cambridge University Press, 60-63. Karp, Herbert H. and K. Dennis Kelly 1971 Toward an Ecological Analysis of Intermetropolitan Migration] Chicago* Markham Publishing Company. Kulldorf, Gunnar 1955 Migration Probabilities. Lund Studies in Geography, Series B. Human Geography, No. 14. Lund* Royal University of Lund, Ladinsky, Jack I967 "Occupational Determinants of Geographic Mobility Among Professional Workers." American Sociological Review 32, 253-264. Lansing, John B. and Eva Mueller 1967 The Geographic Mobility of Labor. Ann Arbor* University of Michigan, Larbor, Gene 1972 "Lagged Response in the Decision to Migrate* A Comment." Journal of Regional Science. Vol. 12, No. 2, 307-32S: Lee, E.S. 1965 "A Theory of Migration." Demography 3, 47-57. Leven, Charles L. 1959 "Population, Migration and Regional Economic Develop ment." Current Economic Comment 21 (November), 31-42. Levy, M.B, and W.J. Wadycki 1972 "Lifetime Versus One-year Migration in Venezuela," Journal of Regional Science. Vol. 12, No. 3, 407-415 Lianos, Theodore P. 1970 "A Stocks and Flows Approach to Migration," Ameri can Journal of Agricultural Economics 52, 442-443. 1971 "Labor Mobility and Market Imperfections." Canadian Journal of Agricultural Economics 18, 97-10F] 13 ^ 1972 "The Migration Process and Time Lags." Journal of Regional Science. Vol. 12, No. 3* 425-^3^ Long, Larry H. 1970 "On Measuring Geographic Mobility." Journal of the American Statistical Association 65» 1195"12°3» 1972 "The Influence of Number and Ages of Children on Residential Mobility." Demography 9» 317-382. 1973 "Migration Differentials by Education and Occupa tions Trends and Variations." Demography. Vol. 10, No. 2 (May), 2*1-3-258. Lowry, I.S. 1966 Migration and Metropolitan Growths Two Analytical Models. San Franciscos Chandler Publishing Company. Mclnnis, R. Marvin 1970 "Provincial Migration and Differential Economic Opportunity." In Migration in Canadas Some Regional Patterns. Edited by L.D. Stone. Ottawas Queens Printer. Mangalam, J.J. and Cornelia Morgan 1968 Human Migrations A Guide to Migration Literature in English 19*55-1962. Lexington, Kentuckys University of Kentucky Press. Miller, Edward 1972 "A Note on the Role of Distance in Migrations Costs of Mobility Versus Intervening Opportunities." Journal of Regional Science, Vol 12, No. 3» *475-^78. Morrison, Peter A. 1967 "Duration of Residence and Prospective Migrations The Evaluation of a Stochastic Model." Demography 553-561. 1973 "How Population Movements Shape National Growth." Paper No. 5007, Rand Corporation (May). Nelson, P. 1959 "Migration, Real Income and Information," Journal of Regional Science 1, *13-73. Okun, B. and R.W. Richardson 1961 "Regional Income Inequality and Internal Population Migration." Economic Development and Cultural Change 9, 128-143. 135 Orange County Health Planning Council 1973 Master Plan of Health Facilities and Servicesi Orange County. 1973. Santa Ana, California* Orange County Health Planning Council. Petersen, William 1958 MA General Typology of Migration." American Socio logical Review 23* 256-266. 1961a "Internal Migration," In Population. New York* Macmillan, 153-178. 1961b "The General Determinants of Migration," In Popu lation. New York* Macmillan, 592-621. Prehn, John W, 1967 "Vertical Mobility and Community Type as Factors in the Migration of College Graduates." Demography 4, 283-292. Rabianski, Joseph 1971 "Real Earnings and Human Migration." Journal of Human Resources. VI*2, 185-192. Raimon, R,L, 1962 "Interstate Migration and Wage Theory," Review of Economics and Statistics 44, 428-438, Ravenstein, E.G. I885 "The Laws of Migration." Journal of the Royal Statistical Society 48, 167-227. I889 "The Laws of Migration." Journal of the Royal Statistical Society 52, 241-305. Rogers, Andrei 1967 "Estimating Interregional Population and Migration Operators from Interregional Population Distri- e1 butions." Demography 4, 515-531. 1973 "Estimating Internal Migration from Incomplete Data Using Model Multiregional Life Tables.” Demography. Vol. 10, No. 2 (May), 277-287. Rose, A.M. 1958 "Distance of Migration and Socio-economic Status of Migrants," American Sociological Review 23 (August), 420-423, Ross, H. Laurence 1962 "Reasons for Moves to and from a Central City Area." Social Forces 40 (March), 261-263. 136 Saben, S, 1964 "Geographic Mobility and Employment Status, March 1962 - March 1963." Monthly Labor Review (August) 870-877. Sahota, G.S, 1968 "An Economic Analysis of Internal Migration in Brazil." Journal of Political Economy 76, 218-245. Schnore, Leo F, 1961 "Social Mobility in Demographic Perspective." American Sociological Review. XXVI (June), 407-423. Schoenberg, Ronald 1972 "Issues in Multiple Regression Revisited." Paper read at the 67th Annual Meeting of the American Sociological Association. Schultz, T.P. 1971 "Rural-urban Migration in Columbia." Review of Economics and Statistics 53* 157-163. Schwartz, A.B, 1971 "On Efficiency of Migration," Journal of Human Resources. VI*2, 193-205. Shryock, Henry S., Jr. 1964 Population Mobility Within the United States. Chicago 1 Community and Family Study Center, Uni versity of Chicago, Shryock, Henry S., Jr., and Charles B. Nam 1965 "Educational Selectivity of Interregional Migra tion." Social Forces 43, 299-310. Sjaastad, L.A. 1960 "The Relationship Between Migration and Income in the United States." Papers and Proceedings of the Regional Science Association 6, 37-64, 1961 "Income and Migration in the United States." Ph.D. dissertation. University of Chicago. 1962 "Costs and Returns of Human Migration." Journal of Political Economy 70 (October, Supplement), 80-93, Southern California Association of Governments 1973 "Population Growth Analysis," An unpublished paper (April 15). 137 Spiegelglas, Stephan 1961 "Role of Industrial Development as a Factor Influencing Migration to and from Wisconsin Counties, 1940-1950•" Journal of Farm Economics 43 (February), 128-137. Stewart, Charles T., Jr. I960 "Migration as a Function of Population and Dis tance,” American Sociological Review 25 (June), 347-356. Stewart J.Q. 1941 "The Gravitational Geographic Drawing Power of a College," Bulletin of the American Association of University Professors. XXVII, 70-75. Stouffer, S.A. 1940 "Intervening Opportunities* A Theory Relating Mobility and Distance." American Sociological Review 5» 845-867. 1960 "Intervening Opportunities and Competing Migrants." Journal of Regional Science 2, 1-26. 1962 Social Research to Test Ideas. New York* Free Press of Glencoe, 66-112, Strodtbeck, F.L, 1949 "Equal Opportunity Intervals* A Contribution to the Method of Intervening Opportunity Analysis." American Sociological Review 14, 490-497, Stub, H,R, 1962 "The Occupational Characteristics of Migrants to Duluth* A Retest of Rose's Hypothesis," American Sociological Review 27 (February), 87-90, Suval, Elizabeth M. and C. Horace Hamilton 1965 "Some New Evidence on Educational Selectivity in Migration to and from the South." Social Forces 43, 536-547. Taeuber, Karl E. 1961 "Duration-of-residence Analysis of Internal Migra tion in the United States." Milbank Memorial Fund Quarterly. XXXIX, 120. Taeuber, Karl E., William Haenszel and Monroe G. Sirken 1961 "Residence Histories and Exposure Residences for the United States Population." Journal of the American Statistical Association. LVI. 824-834. 138 Tarver, James D, 1961 "Predicting Migration." Social Forces 39 (March), 207-214, 1965 "Metropolitan Area Intercounty Migration Rates* A Test of Labor Market Theory." Industrial and Labor Relations Review 18, 213-223. Tarver, James D. and R. Douglas McLeod 1970 "Trends in Distances Moved by Interstate Migrants." Rural Sociology 35, 523-533. 1973 "A Test and Modification of Zipf's Hypothesis for Predicting Interstate Migration," Demography. IV*2 (May), 259-275. ter Heide, H. 1963 "Migration Models and Their Significance for Popu lation Forecasts." Milbank Memorial Fund Quarterly. XLI, 56-76. Thomas, Dorothy S, 1958 "Age and Economic Differentials in Interstate Migration." Population Index 24, 313-325* Thomlinson, Ralph 1961 "A Model for Migration Analysis." Journal of the American Statistical Association 56# 675-686. 1962 "The Determination of a Base Population for Com puting Migration Rates." The Milbank Memorial Fund Quarterly 15, 356-366. Vanderkamp, John 1968 "Interregional Mobility in Canada* A Study of the Time Pattern of Migration." Canadian Journal of Economics 1 (August), 595-608, 1971 "Migration Flows, the Determinants and the Effects on Return Migration," Journal of Political Economy 79» 1012-1031. Wakeley, Ray E. and Mohiey Eldin Nasrat 1961 "Sociological Analysis of Population Migration." Rural Sociology 26 (March), 15-23. Wilber, George W. 1963 "Migration Expectancy in the United States." Journal of the American Statistical Association 58, 444-453. 139 Woofter T J 1967 '"Migration in the Southeast." Demography 4, 532- 552. Zachariah, K.C. 1966 "Bombay Migration Studyt A Pilot Analysis of Migra tion to an ABian Metropolis." Demography 3, 378“ 392. Zipf, G.K. 1946 "The P,P2/D Hypothesis on the Intercity Movement of Persons," American Sociological Review. XI, 677-686.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Determinants Of Neonatal And Postneonatal Mortality: The Case Of Malawi, 1992

PDF

Differential Fertility Behavior And Values In Rural And Semi-Urban Costa Rica

PDF

Economic Differentiation And Social Organization Of Standard Metropolitanareas In The United States: 1950

PDF

Referential Dissociation And Response To Stress

PDF

The Construction And Empirical Test Of A Theory Based On Selected Variables In Small-Group Interaction

PDF

Ecology Of Negro Communities In Los Angeles County: 1940-1959

PDF

The Effects Of Social Security Programs On Fertility Levels

PDF

An Empirical Examination Of The Relationship Of Vertical Occupational Mobility And Horizontal Residential Mobility

PDF

Social Components Of Housing Cost In The Western Metropolis

PDF

Status Consistency Among The Clergy

PDF

Familism, Suburbanization, And Residential Mobility In A Metropolis

PDF

Employment Distribution, Income, And City Size: A Statistical Analysis

PDF

Consensus Of Role Perceptions In A Welfare Planning Council

PDF

Prediction Of Overt Behaviors In Hospitalized Psychiatric Patients

PDF

Transformational Processing Of Sentences Containing Adjectival Modifiers

PDF

The Attitude Of Resistance To The Adoption Of An Older Child

PDF

Structural Factors Affecting Fertility In Large United States Cities

PDF

Conflict And Generalized Conflict In Verbal And Motor Responses

PDF

Delinquency And Rationalization: A Study Of The Delinquent Act

PDF

The Selfish Self: A Social Psychological Study Of Social Character

Asset Metadata

Creator Berg, Dennis Floyd (author)

Core Title Determinants Of Intercounty Migration: California, 1970-1973

Degree Doctor of Philosophy

Degree Program Sociology

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag OAI-PMH Harvest,sociology, demography

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Van Arsdol, Maurice D., Jr. (committee chair), Gutman, Jonathan (committee member), Heer, David M. (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c18-719989

Unique identifier UC11356480

Identifier 7426014.pdf (filename),usctheses-c18-719989 (legacy record id)

Legacy Identifier 7426014.pdf

Dmrecord 719989

Document Type Dissertation

Rights Berg, Dennis Floyd

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA